Sometimes we can give the type to MRE::take_safely

[ggevay]

More typ() call elimination, similar to e.g. #30878

This one goes after various take_safely calls, where we know the type already, so no need to call .typ().

Also, this makes take_safely() give a better key and nullability to the constant, making use of the fact that we have an empty collection, so we have the keys [[]] and nothing is nullable.

Motivation

• This PR fixes a previously unreported bug: take_safely was often not giving the best possible key to the constant. Ideally, the keys that are recorded on a MRE::Constant should be the same as if we were to call keys_with_input_keys on it, so that we don't have anomalies such as create_fast_path_plan needing to // For best accuracy, we need to recalculate typ.. This PR moves one step closer to this.

• This PR refactors existing code.

Tips for reviewer

Checklist

• This PR has adequate test coverage / QA involvement has been duly considered. (trigger-ci for additional test/nightly runs)
• This PR has an associated up-to-date design doc, is a design doc (template), or is sufficiently small to not require a design.
• If this PR evolves an existing $T ⇔ Proto$T mapping (possibly in a backwards-incompatible way), then it is tagged with a T-proto label.
• If this PR will require changes to cloud orchestration or tests, there is a companion cloud PR to account for those changes that is tagged with the release-blocker label (example).
• If this PR includes major user-facing behavior changes, I have pinged the relevant PM to schedule a changelog post.

[ggevay]

Tagging @frankmcsherry because of touching EQProp.

[ggevay]

Also, maybe this eliminates that key derivation which @antiguru found to be a bottleneck in FoldConstants? Which one was that?

[ggevay]

Bringing back to draft, because there are test failures.

[ggevay]

Fixed

[mgree]

Seems fine to me!

One possible change, if we're going golfing: you could have take_safely_with_rel_type take an Option<RelationType>, saving us the double call to MRE::typ() when calling MRE::take_safely(). Only matters in debug mode, so I'm not sure it's worth it.