Changed Any Operator to be less Wasteful

[jff-work]

As suggested in issue:
#1552

The Any operator in Quint was evaluating a lot of statements in cases that could be skipped. I implemented the proposed solution and wrote some tests as well as reasoned about the probabilities of different statements being returned by Any, which remain the same for this implementation.
https://github.com/jff-work/quint-any-test

I'd probably run some more complicated tests before merging this.

[bugarela]

Hi! Thanks a lot for addressing this and documenting everything so well. I'll do some testing today - the failing tests don't mean there's anything wrong, but this does (as expected) change the way rand() is used, and therefore runs with a fixed seed will have different results under this new implementation. I just need to update the seeds in the failing tests.

Sorry for the delayed response, I was on vacation for a while. I'll get back soon with a proper review.

[bugarela]

I just tried this change on the (private) project where I noticed the slowdown caused by any and, for this instance, this change is a 2x speed up 🎉

[bugarela]

Looks great! I've updated the integration tests that relied on the seed and tested this on a couple of real projects, all seems great and the performance improved so much!