The CHA limit per long rest was to prevent higher level sorcerers from routinely adding 1st and even 2nd level spells throughout the day. It might be overkill since I imagine more sorcerers a likely to hoard sorcerer points for metamagic. But, when you consider that you can swap a spell slot for additional metamagic, it might result in too many uses per long rest.
Also, since CHA represents inner strength or whatever, tapping into that inner wellspring of magic only a few times makes sense to me flavor-wise as well.
We can play test it with the limit in place, and if we find it is in greater demand (I don't think that will be the case), we can switch it to a short or long rest, or remove it entirely.
The question is not what the CHA limit is hedging against.
Hedging against issues one imagines might arise at high level on the first prototype of a rule is not the most effective approach.
@FrogReaver So far as I can tell, that is
exactly why UA material is typically over, rather than under, powered. It's not a flaw, it is an approach that professional designers use to their advantage. Big caveat: it's homebrew so of course one should do as one enjoys. I'm just chiming in to offer tools that could help enhance what people can achieve.
So if you think about what would be (and is) a methodology likely to consistently produce strong mechanics that play well. There will need to be problems or goals to solve for (here we have a great example), divergent solutions, and then prototypes of solutions. The most likely looking prototypes will need to be tested in a range of circumstances, although the focus will be the way the game is mostly played.
A really effective technique is to look at a mechanic that captures as simply as possible your intent, with minimum trimming. It's hard to do, and usually you over-design to start with. The trimming is almost always a layer of obfuscation, that clouds what you could achieve.
Think then about
@dnd4vr's speculation. Maybe he is exactly on the money? How would one even know if one only tests a CHA/rest limited version? The best outcome he can get from playtests is to learn if CHA/rest limit comes into play sooner than the sorcery points limit does (meaning, ironically, that it is unwarranted). Or he will discover that the CHA/rest limit does kick-in before SP runs out and then...? He still won't know if allowing more was going to be egregious.
And think about
@FrogReaver's speculation (2), which strikes out in the opposite direction. The overwhelming majority of opinion I have read about SP is that they are too few, not too many. So I agree with
@FrogReaver. Even if I did not, however, I don't think you prove this as quickly as possible by double-braking.
And that is the point, really. Playtesting is a scarce resource. It's hard enough to get the group together, explain the rule, try it out in
only one circumstance! Let alone do it for the necessary canon. Playtesting also aligns with better game mechanics: there's a strong correlation between more testing and better games. One needs to make each bout of testing as effective as possible. Any obfuscation introduced into the solution just gets in the way of quickly judging how well your core idea has solved the problem.