To get a grip on how much time Thaumaturgy takes, you'll want to skip to YS p. 268, "Adjudicating Preparation". Also note that if their Lore is sufficient, you can consider the wizard already prepared and it takes no more play time than an Evocation.
The significant unit of time for Thaumatury isn't a minute, or second, or even hour. Thaumaturgy isn't measured in game-world time units, but rather in game-play time units. Hence, the significant unit of time is the scene, and the amount of time it takes is a matter of GM pacing.
In essence, roleplay the preparation. Do they know what the ritual they need to do is? Their character has to actually do those things. Do they not know the ritual they need? Then they have to go find out and gather the knowledge and resources. Play that out as a quick montage, possibly with skill checks.
That's what the options for the player are there for: Invoke Aspects, Make Declarations, Accept or Inflict Consequences, and (for extreme cases) Skip Scenes.
None of those things should be done as a simple roll and move on. Play it. Playing it out is where the time comes from. Do they have a bloody handkerchief of the person they're trying to locate? That's great, but they still have to do all the FATE stuff to make it mechanically relevant, and to make it mechanically relevant they have to make it roleplaying-ly relevant by playing how it matters and how they use it. This is the beating heart of FATE, where the mechanics demand that roleplaying happen before you can use them. Ignoring this makes FATE fall flat. FATE is a game about showing why the mechanics are justified by playing them first, not by just telling the GM what mechanics are being used and skipping straight to their mechanical effect.
For some examples of what this looks like:
Make a declaration to create an Aspect about it. What are they doing with the handkerchief? Are they focusing their will on it, attuning themselves to the substance of the person they're trying to find? Are they using it to create the circle? Are they arranging it "just so" with a few other personal articles to create a "presence" of the person? Are they smearing symbols on a knife they're going to transform into a "compass"? Ask them what they're doing! What they do will determine what they roll to create the Aspect and what the Aspect will be. Don't just give it to them – make them earn it. Without doing that, you'll skip what makes magic in FATE actually fun to use.
Invoke Aspects. Same idea: they have to earn it by playing it. Players can't just say, "I'm invoking Dear to My Heart because I care about her a lot", they have to show it. They're going to have to roleplay some heart-rending stuff if they want that Aspect. It doesn't have to be long – in fact it can be very brief, as brief as a sentence. But it has to be there, or they don't get the Aspect. Put the fiction in place first, and the mechanics that belong will always be obvious.
Accept or Inflict consequences. In Storm Front, Harry didn't just say, "So I'm going to channel the storm now," and then he cast a spell at the demon. No, he agonised over exactly how dangerous that was, and then did it anyway. He didn't just vaguely channel the storm, he got hit by freaking lightning. He scorched and burned and ached. That is what your players need to do when they're accepting a consequence. The example in the book of the nasty ways consequences can be inflicted takes time too: torturing someone, or sacrificing a living victim, aren't instantaneous. The dark wizard is going to have to get a hold of them, bind them, and actually do the deed. You as the GM need to make that a scene, perhaps a pivotal one. That is where the casting time comes from.
Skip Scenes. This is the fall-back. It's actually the least interesting option, because it takes the player out of the scenes the others are part of. They still need to describe what's going on, else they don't get this option. It should never be: "Okay, I skip a scene to get more shifts." Even the book makes this clear: The player has to describe, even if only briefly, what they're doing that makes them have to skip a scene. Research in the library, ransacking their lab for that one component that they need, and so on.
It should be clear now where the time that Thaumaturgy takes comes from. It comes from all the stuff that the wizard has to do to actually pull off a complex spell, all that stuff that needs to be said by the player and done by the PC. It can sometimes be done as a montage: You don't have to describe every step to the library, just a sentence that evokes images of riffling through shelves, poring over books, and burning candles late into the night – stuff that shows time is passing, and why it's passing.
This is also why Thaumaturgy isn't always strictly better than Evocation: if the character has to actually do this stuff, the world isn't always going to wait for them. If they have to 1) declare an aspect, 2) invoke that and a few other aspects, 3) draw the ritual circle, 4) do the actual ritual, then at some point in that multiple-action process the Bad Guys are almost certainly going to have opportunities to cause problems. It might not even be direct interference – action that the player doesn't like might continue elsewhere.
If a PC wizard is casting Thaumaturgy in the middle of period where time is of the essence – an action scene, or when time is running out some other way – take this golden opportunity to make their life very, very interesting while they're distracted. Interact with their preparations and ritual. They run into exactly the person they want to avoid while trying to get to the library. They need to draw a circle but they only have chalk and it's raining – now what do they do instead? Do they shrug and forego the shift they could get, or do they scramble for an alternative way to create the circle?
Make it fun, make it interesting, play it out, and the time that Thaumaturgy takes will happen all by itself without counting it out.
So far, we have two answers, which appear to contradict each other. I tend to think in practice they're not that far apart from each other, tho.
When you're taken out, you cede control over your fate to the attacker. That means the attacker can assert all sorts of things about what happens to you. Like: you're dead. And because total destruction of the character is on the table, so is any amount of change to the character. Like: you live, with all your consequence slots filled and needing healing. Or: I just knock you out, you'll have a bruise and maybe a headache for the next scene or so — let's call that a mild consequence (or not even that).
(And that's just assuming a physical conflict. In other contexts taken out might be, "And so I utterly change your view of your father. Change the aspect representing your relationship with him on your sheet to reflect the resulting estrangement.")
At any rate, all that nastiness that could befall you (consequences, etc) is still in play after you're taken out. So any notion of saying "sure, I'm taken out" as a way of avoiding consequences is bunk; the only way to avoid consequences for sure is to concede when it's properly time to concede.
So: how much does it matter if the dice hit the table, you take some stress and consequences, and then immediately concede (as has been said) vs giving up and having stuff at least as bad happen to you anyway? Not too much, IMO.
But, anyway, let's look at page 140: It's a summary of the effect, so you can quickly understand what success on an attack means; it's not procedure. There's a page reference to stress and consequences in the sidebar, thereby implying that the official procedure is found elsewhere: page 160.
Doppelgreener's on the mark by taking this text to heart: "If you get hit by an attack, one of two things happen: either you absorb the hit and stay in the fight, or you’re taken out."
As you read on through page 160-162, stress and consequences are referred to as options. Options imply choice. For me, that makes the procedure (essentially): look at value of hit; choose whether or not to take consequences to reduce it; look at post-consequence value; mark off appropriate stress box; if you can't (or won't, tho given that stress is super-ephemeral that'd be a weird line to draw in the sand) mark off the appropriate stress box, you're taken out.
What's being "forced", really, is the handling of the hit, with concession unavailable until after you handle it. Dice are on the table, that hit's coming your way, and you've gotta parse it out with your options or drop, and if you drop, you have no control over what happens to you. Choose to take consequences, and you'll have some control over what's happening to you (including, to a reasonable extent, how those consequences are described); choose not to take consequences and leave yourself with more stress than you can actually take, and you give up all control as you're taken out... including control over whether or not you end up with consequences.
As such, yeah, you could choose not to reduce an incoming hit with consequences (that's the real crux of all of this) and just get taken out by the resulting massive pile of stress. Because it doesn't really matter that you chose not to take those consequences. When you're taken out anything can be done to you: "the person who took you out gets to decide what your loss looks like and what happens to you after the conflict", page 168. I can't see how there's any value in forcing someone to retain control over their character when they don't want it, which is what choosing to be taken out (instead of conceding) is.
Apologies if this seems confused or rambly — it's late!
Best Answer
It takes as many shifts of harm to take someone out as it takes. The amount is arbitrary, since the character may have more or less stress boxes, consequences, stunts, etc. Asking how many is kinda like asking how long a piece of string is.
To calculate someone's plot armor against getting taken out by a single attack, you'll need to add up the following, assuming they're willing to spend everything they can on avoiding being taken out:
Assuming all but the first two points add to nil, this results in a minimum of 23-25 shifts (20 from consequences + 2-4 from a stress box + 1) to take out a player character in one blow, if they have no extra consequences. This excludes their defence roll, which will add ±4 and their defence skill, too.
If you don't quite manage to take them out (e.g. you only get 20), the character will still be pretty hobbled after this, and significantly changed by the experience of whatever happened to them in this attack, thanks to their extreme consequence!
In a practical situation, this is a lot of variability, so if you ask me how many shifts a character can take, I'll be asking you to show me the character and circumstances.
Bear in mind, though, this doesn't make the system deadly. The majority of the section on getting taken out is on character death, and basically says it's boring. Even if someone's got a shotgun aimed at them, they probably won't be taken out by actually being hit by it.
Further, you mentioned knocking someone out via consequences - that's probably the least reliable way to knock them out. You're probably better off using create an advantage to place a knocked out aspect on them.