This answer includes a frame challenge.
0. Let go of the fallacy of perfect balance
What I am looking for is a good resolution, or compromise, so that my
players don't feel slighted and no one ends up with very high or very
low numbers, creating overshadowing or spotlighting issues.
Oddly enough, over the years that I played "roll 'em up" since 1975, we somehow managed to have fun without point buy. For decades. Loads of fun, with some characters a bit stronger than others. (Heck, even with point buy, please see "tiers" in 3.5 to see how "balance" can still be an issue).
Current experience: we rolled our characters in the 5e campaign I am in now. No point buy. I've known all of these guys since high school, except my nephew and the DM. That's since about mid 1970's. Differing personalities. Rolling is no obstacle to fun and success.
If all of you attend this game with "let's have fun" as the goal, then let's get to addressing your problem.
As stated, your problem isn't dice rolling or not, per the email extract you provided us. Your problem is the relationship between players and DM.
What problem should you solve first? "Player versus DM attitude."
Find Common Ground
Before session one, you need a face to face session with everybody who will play to iron out what you expect from the game, and what they expect. Listen, and then share what you expect and why you think point buy is a good idea. Then, as a full table of people, come to a consensus.
Rolling Doesn't Hurt
Rolling for stats is a 40 year old feature of RPG games that works well enough. If that is what all of your players want, where's the harm in letting them do that? Your forecasting drama and problems is not giving anyone, yourself included, the benefit of the doubt. You will all be able to have fun, regardless. Anyone griping later on gets "the look" and the following response: "you all agreed to roll for stats, let's press on with the game, make the best of the tools you have." Since you mentioned that this is the gaming group you've been with for 2.5 years, and you've been friends with some of them for about 15 years, you all already know how to have fun together. Have some faith in that already established track record.
If the opinion is divided among the players on this point, go back to point one and the pre-meeting to iron out expectations before character creation. Until you reach consensus, the game is at risk.
Play!
Once expectations are more or less aligned, play and have fun.
Have Fun!
If you aren't having fun, you are doing something wrong (as a group). That would be the topic of an entirely different question, after you all begin. I see this question as "an ounce of prevention" effort (a good idea as you know this group). Given that you've known each other a while, up to a long while, I'd be very surprised if you don't have fun if you start on the right foot. Consensus building is a means to that end that's pretty effective.
On a mechanical note, since this is 5e, an ASI (later) or a racial bonus (at start) can take the hard edge off of a low number.
Dual wielding daggers does scale way better with your sneak attack than using only one rapier.
At first if all attacks hit the damage calculation would be:
$$
\text{Two Daggers}: 2d4+\text{DEX}+\text{SNEAKATK} \\
\text{Rapier}: 1d8+\text{DEX}+\text{SNEAKATK}
$$
With Dex = 3 and Sneakatk = 1d6 that would be:
$$
\text{Two Daggers}: 6-17 \text{(Avg. 11.5)} \\
\text{Rapier}: 5-17 \text{(Avg. 11)}
$$
Variable Hit Chances:
The problem is, that not all attacks hit.
We know that daggers and rapiers use the same attack bonus, thus having exactly the same chance to hit.
$$
\text{Two Daggers}= \text{Hit Chance}\times(1d4+3)+\text{Hit Chance}\times(1d4)+\text{SNEAKATK}\\
\text{Rapier}= \text{Hit Chance}\times(1d8+3)+\text{SNEAKATK}
$$
Calculation for Sneak Attack:
If we assume, that every hit is eligible for sneak attack, we can add the damage to the daggers if any of the two attacks hit. If my math does not fail me that calculates as follows:
(For two Daggers)
$$
\begin{align}
\text{One Attack Misses} &= (1-\text{Hit Chance}) \\
\text{Both Attack Miss} &= (\text{One Attack Misses})^2 \\
\text{Sneak Attack Chance (NOT Both Attack Miss)} &= 1-(\text{Both Attack Miss}) \\
&= 1-(1-\text{Hit Chance})^2
\end{align}
$$
Also let us talk in average damage for ease:
$$
\text{Two Daggers}: \text{Hit Chance}\times(8)+(1-(1-\text{Hit Chance})^2)\times(3.5) \\ \text{Rapier}: \text{Hit Chance}\times(7.5) + \text{Hit Chance}\times(3.5)
$$
Base Damage:
Regardless of HitChance the daggers deal more damage in average. As the HitChance can never be less than 0.05:
$$ \text{Hit Chance}\times8 > \text{Hit Chance}\times7.5 $$
Sneak Attack damage:
The damage would be equal in both cases, so the relevant part is the trigger chance. Assuming the attacks are eligible for a Sneak Attack, the relevant part would be whether you hit or not.
When attacking twice, your chance to hit is higher than when only attacking once.
$$ (1-(1-\text{Hit Chance})^2) \geq \text{Hit Chance} $$
Assuming 0 < HitChance ≤ 1
Proof (Thanks to @Glen_b):
$$
1-(1-H)^2 = 1-(1-2H+H^2) = 2H-H^2 \\
= H+H(1-H) \geq H \forall 0\leq H \leq 1
$$
with equality only possible at the endpoints and a maximum difference at $$H=0.5$$
Here is a graph to visualize the chances that you can apply your sneak attack damage:
Y = Sneak Attack Chance; X = Hit Chance per Attack; Red = Two Daggers; Blue = Rapier
Thus, two daggers are better with and without sneak attack.
Here is another graph visualizing the average damage for both weapons (Assuming +3 Dexterity modifier, 1d6 sneak attack, no magical bonuses, all attacks eligible for sneak attack.)
Y = Damage; X = Hit Chance per Attack; Red = Two Daggers; Blue = Rapier
Opportunity Costs
Through OPs comment on another answer, I know his rogue is level 2.
This means he could use his bonus action to Dash, Disengage or Hide.
Later on, depending on your subclass and table rules, you might gain Feats or other class features like Fast Hands which require your bonus action. So, why would I take the Two Daggers instead of the Rapier? This would trade in positioning possibilities or other maybe useful opportunities for a measly 0.5 damage.
While the damage difference might appear to be 0.5 in average, the real advantage of dual wielding comes with the sneak attack chance.
Look at it this way:
If you have only 50% hit chance the difference would be
$$ (0.5\times8+0.75\times3.5) - (0.5\times(7.5+3.5)) = 1.125 $$
The daggers now deal 1.125 damage more in average.
Same scenario, but without sneak attack:
$$ (0.5\times8)-(0.5\times7.5) = 0.25 $$
Only a 0.25 damage difference. This gap increases drastically as sneak attack damage increases.
So your goal is to get your sneak attack damage through. If your first attack hits, your rapier deals in average 2 damage more than the dagger, but if the first attack misses you deal no damage with the rapier, while dual wielding gives you another attack to possibly deal your sneak attack damage.
Why is this so important?
When looking at your weapon attack damage, how does it scale?
- Your Dexterity modifier might go up (but would benefit both weapons equally)
- Your sneak attack damage increases with your level (better for daggers, because of higher hit chance if you attack twice)
- You may get a magic weapon
The 2 Avg. damage difference from the damage dice, if the first attack hits looses its relevance while the increased chance of sneak attack gains in relevance.
Let's look at a level 5 rogue who used is ASI on Dex (+4) and has now 3d6 Sneak attack damage. With a 50% hit chance: [the numbers in square brackets are without sneak attack]
$$
\begin{align}
\text{One Dagger: } 0.5\times(2.5+4)+0.5\times(10.5) = 8.5 \text{ [3.25]}\\
\text{Two Daggers: } 0.5\times(2.5+4)+0.5\times(2.5)+0.75\times(10.5) = 12.375 \text{ [4.5]} \\
\text{Rapier: } 0.5\times(4.5+4)+0.5\times(10.5) = 9.5 \text{ [4.25]} \\
\text{Rapier+1: } 0.55\times(4.5+4+1)+0.55\times(10.5) = 11 \text{ [5.225]}
\end{align}
$$
Y = Damage; X = Hit Chance per Attack; Red = Two Daggers; Blue = Rapier
Even a Rapier+1 does not reach the Avg. damage of two daggers.
Here an level 20 (+5 Dex, 10d6 Sneak attack damage) example:
Y = Damage; X = Hit Chance per Attack; Red = Two Daggers; Blue = Rapier
The damage gap increases, but to get the most out of your turn I would recommend the following:
If your first attack with dual daggers hit, accept that you might have dealt more damage with a rapier and use your bonus action to get away (improve your survivability). If this is no option you can attack another time to maybe deal 1-4 extra damage.
If the first attack misses, attack another time, this is where dual wielding shines.
Best Answer
Your Constitution modifier can decrease your starting hit points
The rules on "Hit Points and Hit Dice" state:
Nowhere does it make any exception for negative Constitution modifiers. In fact, I don't know if there is any feature where you add a modifier only if it is positive.
Instead of ignoring negative modifiers, feature often impose minimums on the total such as when gaining health at level other than 1:
Notably, a negative Constitution modifier still applies it just also gets effectively ignored if the modifier is sufficiently negative to change the total to a non-positive number.
Applying this sort of minimum to your starting hit points would be awkward to phrase, getting you something like "minimum of the maximum of your hit die", which clearly isn't the best wording (though that doesn't particularly mean much).
Because nothing explicitly states that a negative Constitution modifier doesn't apply and nothing states a minimum on your starting hit points, your Constitution modifier can actually decrease your starting hit points.
A possible example of what the section could have said
There might actually at least one feature that specifies a minimum on the modifier to a roll. The Artillerist Artificer's Protector Eldritch Canon states:
It looks like this minimum applies to the modifier and not the total, though perhaps this feature is just weirdly worded and has a "+" to denote positivity for some reason... Regardless, such a wording does not appear in the section on starting hit points.