Seedable hashmap for testing?

[workingjubilee]

This is inspired by the many regressions in #128899 that appeared to be "sorting" issues but actually seem to be hashmap issues. It would be nice if we had some way, even if only for unstable/nightly/internal/literally-just-crater usage, to tweak std's hashmap (or just the RNG for it?) so that we could determine if the hashmap's iteration order is the reason a crate's tests are failing.

[ChrisDenton]

Couldn't this be done with a PR that simply tweaks RandomState and does a crater?

[workingjubilee]

@ChrisDenton Doesn't work if we want to control both "before" and "after" of RandomState.