fix(memory): recover stale LRU high-priority weight

What's changed and what's your intention?

Fix LRU high-priority weight recovery when the tracked counter drifts from the intrusive high-priority list.

The overflow path assumed high_priority_weight > high_priority_weight_capacity always implied a non-empty high-priority list, and could panic if the accounting was already stale. This PR repairs stale accounting by recomputing the actual high-priority list weight when recovery is needed, and routes high-priority decrements through a checked helper used by overflow, pop, remove, and acquire paths.

A user of a crate I maintain ran into this scenario in nntp-proxy#73.

Checklist

• I have written the necessary rustdoc comments
• I have added the necessary unit tests and integration tests
• I have passed cargo x (or cargo x --fast instead if the old tests are not modified) in my local environment.

cargo x --fast was run; clippy, strict test run, doctests, docs, examples, and cargo-machete passed, then the final legacy license-eye header check step failed. The current license workflow command, hawkeye check, passes.