Four lesser options, one major one.
Fast-forward the split party part. Don't dwell on stuff as much as you otherwise would.
Snack and bathroom break time. Shoo the other players off. This will serve as a cue to the sneakers that they should include everyone eventually.
Metagame, encourage the group to find ways to include people even when it's suboptimal.
Spring stuff on the other party to keep them occupied. Sure, if they just get a lead, they can wait to follow up, so make the action come to the PCs - someone kicks down the door or otherwise looks to interact with them NOW.
Give them all equal spotlight time. 5 minutes dealing with sneakers or interrogators, 5 with the others. They'll either start going to do stuff of their own accord, or they will, heaven forfend, roleplay with each other.
Number 5 is really my favorite. We've had plenty of great moments while half the party is doing something "important" and the other half goes to a bar and gets in trouble. All you have to do is give them the spotlight time, PCs are notoriously twitchy and impatient and will find ways to entertain themselves, "on task" or not.
My added thoughts based on your question edit.
You don't need to balance "in game" time. If one group went and did something for six hours, but real time that took five minutes, then you just need five minutes of spotlight time for the other group, even if that's five minutes of rousting a hobo and then 5:55 of "we go somewhere and hang out."
If they need ideas for what to do or you want to spring things on them - that's why God created random encounters, right? Either they self-motivate some mission oriented stuff, which you come up with, or they mess around in a tavern or go shopping, which you come up with, or they sit around looking glum till a land shark attacks them, which you come up with. You do anything you'd normally do to the full party, but ideally with slightly less kill. (Or not, if you want to dissuade them from splitting in the future.)
If they are content to wait in the bar, let them wait in the bar. Have fun bar things happen. People hang out in bars in real life, it's fun. And sometimes good and/or bad things happen, hence the larger than usual incidence of hookups and cop interventions in bars. All you need is for them to all be having fun and getting roughly equal "spotlight time" (time they get to actually do something at the game table).
Not everything has to be mission related or be a "subplot." They can just find out interesting things about the world they're in. If every story everyone ever tells them is a "subplot," then you get into a bad rut where their expectation is that anything that happens in the game world is FRAUGHT WITH MEANING, and it can't just be some guy BSing about how he fought a troll armed with a spoon once.
In my campaign, I make sure there's a healthy amount of "the world doesn't revolve around the PCs" stuff going on. Not only does it make for a realistic feel, but then when someone wants something to do, they have the expectation that "Hey, I can just go out there and go shopping, or find a bar, or find something to do - I don't have to be 'working through a plot element'" every damn minute of every damn day.
In my Pathfinder campaign recently, the party split. One half was going to do something on task and important. I don't even remember what it was and they probably don't either. The plan was to meet up at a known bar later that day. The other half of the party decided "we'll just go wait at the bar now." I roll a random encounter - giant cockroaches. They get to the bar, and the owner, "Ball-less Bill," an old ex-pirate with one leg, one hand, one eye, and apparently less then one thing down below, was standing outside the bar holding the door shut. "What's the problem?" "My basement flooded and there's these big ol' cockroaches running around in there! They're as big as dogs!" "We can take care of this! Stand aside, Bill!" The two PCs bust in and the cleric casts Call Lightning. The cockroaches are like CR 1/4, so they run all around and out into the street as the cleric blasts lightning holes in everything in sight. The fear of God is put into the local yokels, and the cleric gets an Infamy Point (the equivalent of a Hero Point in my pirate-oriented campaign). They yell "Wooo!" and go in and drink for free.
The other half of the party... They did something on mission. Got some information or something. Who cares. No one remembers that, but they remember the Great Roach Holocaust.
If you have proactive players, this shouldn't be a problem. If you have reactive players, then don't treat them like proactive players and give them 'things to follow up on.' Have the world turn, and its events happen around or to the players. If they decide to just go find another bar rather than chase out the giant cockroaches, fine, then make them choose whether they're drinking "bloodwine or dwarven grog" and go back to team A. Seems to me like you're overthinking this by requiring all events to be part of some big Mythic Plotweb.
Magic items are more interesting when they serve to advance or drive a plot. This may be because
Only with that item can the characters succeed (and it is hard to get or keep the item). Example: wights that can only be hurt by +1 weapons, in an area where +1 weapons are extremely rare.
The item is complex; it only sometimes works, or requires mastery and exploration. Example: a Wand of the Moon which has greater effects at night than during the day, but the nature of the effects change with the phase of the moon (all of which the characters have to find out).
The item has a long and elaborate history, some of which is relevant to some grand plot or event. Example: a horcrux.
The item is unique, and even if not all that powerful, the characters become known as the ones with that unique capability, if only they can make it. Example: the characters have to create +1 swords for wight-slaying, and they learn from an old magician how to both enchant the swords and make them glow if wights are within 100m.
You cannot, in general, make people super-excited about items that are commonplace and easy to get, no matter how fantastic they are. For example, in a matter of seconds, I can talk to any person I choose on the planet, thanks to a cell phone. Do I spend much time thinking about the cell phone itself? Not really. (Apple does a good job at making its products seem extraordinary, so if I had an iPhone it might be a partial exception.)
Best Answer
As with almost anything in roleplay, a little research on the real world can pay large dividends. Begin by looking at real-world police technique.
(To begin within, Google "Reid technique" - badly outdated but still generally used in the US, and appropriate for most fantasy games as well - and the "PEACE method", a more recent approach widely used in Europe.)
To look at your points in order:
The first thing to bear in mind is that an interrogation is not an event in itself, but part of an investigation. We interrogate to get relevant information. So the interrogation is affected by what other clues and information the players have.
Give them leads that they can confirm, or refute, by interrogation. Knowing what questions to ask is half the battle. Let them take that information out of the interrogation, and follow up on it, and come back when they know what else to ask.
Yes, you can and should involve multiple PCs. ("Good cop / bad cop" is one of the oldest interrogation tricks in the world, because it works.)
(They don't all need to be good at interrogation! There's good roleplay in the scene where the subject was about to reveal something, until a cack-handed amaterur lets slip the lead interrogator's bluff...)
Non-interpersonal skills can, and should, provide information that can then be used (via interpersonal skills) to persuade, taunt, inform or lie to the subject. (e.g. Knowing what poison was used on a murder victim may let you intimidate a suspect who could have supplied it... or catch him in a lie.)
If you don't want the interrogation to be an infinite loop, don't loop. People who genuinely "won't talk at all" are extremely rare... and mostly hardened interviewees. Instead of deciding whether the NPC will or won't talk, you should be decided which topics he wants to talk about, and which he is avoiding.
Your "no" can then become a "no, and..." - an attempt by the NPC to steer the conversation in a new direction.
If the PCs fail to get answers, you should have provided some alternate means for information to be obtained and the plot to continue - maybe at a cost in effort, resources, or reliability. For example, the subject might talk willingly... if a favour is done for him first.
(This falls firmly under the "never call for a skill roll you don't want to fail" principle - if a failed interrogation derails your game, why allow it to happen? Either provide an alternate way to progress the plot when it fails, or make sure it succeeds.)
Most things that apply to negotiation apply to interrogation. However, bear in mind that a detective is a bit of a con artist: a negotiation has to think about the subject's opinions afterwards, but an interrogator is (usually) only concerned with the subject's reactions during the invesetigation. An interrogator can afford to lie or bluff to get the desired results.
I recommend that you take a look at Mutant City Blues; being a police-based game, it has a few pages (p109-111) summarising interrogation technique, with GM advice on ways to make it work in play. (It also has a lot of useful ideas about investigation-led game design in general.)