Fix default check combinations

[mmagician]

Description

The trait-default implementation of check_combinations was incorrect. This was not detected because all of the implemented schemes have a custom implementation of this method. We are going to use the default method in Ligero though.

Description of the bug(s):

1. The key in poly_evals: BTreeMap was of type (query_label, point), whereas the first element in the key tuple should have been a poly_label. This was causing map access here to fail with MissingEvaluation.

• poly_query_set was keyed only with poly_label, but held multiple query values with the same poly_label. Iterating over its keys was be done in an arbitrary order.
• The array we zip it with, evals, is returned calling values() on BTreeMap, the values are returned sorted by key of type (poly_label, point).
• So the two constructed arrays were potentially in non-matching orders. To address this, we map the iterated poly labels as (poly_label, point), and collect first, so before it gets combined with a sorted vector of evaluations, it will itself get sorted on the correct key (poly_label, point).
  (2 might seem obvious in retrospect)

As mentioned, this is impossible to validate against the current codebase where all PCS have a custom implementation. To showcase the bug, I've sandwiched the actual fix in between 2 commits, where I first remove the custom impl from marlin_pc and later add it back after the fix. To verify the fix, you can check out 57ac1ab, and run cargo test marlin::marlin_pc::tests::full_end_to_end_equation_test -> should fail.
Then checkout the fix commit dcffaec, run the same command, and verify the tests pass now.

---

Before we can merge this PR, please make sure that all the following items have been
checked off. If any of the checklist items are not applicable, please leave them but
write a little note why.

• Targeted PR against correct branch (master)
• Linked to Github issue with discussion and accepted design OR have an explanation in the PR that describes this work.
• Wrote unit tests
• Updated relevant documentation in the code
• Added a relevant changelog entry to the Pending section in CHANGELOG.md
• Re-reviewed Files changed in the Github PR explorer

[mmagician]

Hmm I might have messed something up, I converted to draft for now

[mmagician]

Ok, I was a bit rash on squashing two loops into one - it turns out we need to first collect before zipping to ensure correct ordering by the new key. Ready now ✅

[Pratyush]

LGTM modulo the nits.

[Pratyush]

By the way @mmagician, in a separate PR, do you mind adding comments explaining what this code is doing? (Since you touched this code most recently you have the best idea.)