Guards: Cache nullGuard predicate.

[aschackmull]

A recent QA run highlighted a performance regression on a single repo: total analysis time went from 3mins to 41mins. I tracked down the cause to a poor join-order in a non-recursive instance of the forex in the wrapperGuard predicate. The predicate is actually recursive, but was being recomputed in a non-recursive setting since its recursion goes through cached dependencies but was (transitively) used in the non-cached nullGuard predicate, which is publicly exposed.

This PR fixes the problem by including the nullGuard predicate in the set of cached predicates in the Guards library.

I could reproduce the performance problem locally and verified that this PR fixed the issue.

Ignore white-space when reviewing to minimize the diff.

[Copilot]

Pull Request Overview

This PR addresses a performance regression where analysis time increased from 3 minutes to 41 minutes by caching the nullGuard predicate. The issue was caused by the nullGuard predicate being used in a non-cached context while transitively depending on cached predicates, leading to poor join-order optimization.

• Moved the ImpliesTC module and nullGuard predicate into a new Cached module
• Added cached annotation to the nullGuard predicate
• Created a public alias for the cached nullGuard predicate

---

_{You can also share your feedback on Copilot code review for a chance to win a $100 gift card. Take the survey.}

[aschackmull]

dca is uneventful