[BasicContainers] Assorted hashed container fixes and improvements

[lorentey]

RigidSet/UniqueSet/RigidDictionary/UniqueDictionary currently have an implementation of insert that sometimes calculates a maximum probe count that is slightly lower than the real value, and when that happens find will call off the search for some keys right before it would have found the match -- resulting in some of the contents being inaccessible.

Implement a thorough _checkInvariants method to verify hash table layout at runtime. Use it in the existing (and some new) RigidSet unit tests to uncover corrupt results as they happen. Debug and fix their root causes.

While we're here, implement a handful of important TODO items and low-hanging performance fruit:

• Behavioral engineering: When inserting into small sets/dictionaries, avoid simply appending the new item to the end of storage: it gives the false impression that these types would be preserving insertion order. Instead, insert the new item near the middle of existing storage, swapping the old value to the end.
• Performance: When inserting an item whose ideal bucket is already occupied, avoid hashing the item that's already there -- it cannot possibly have a lower probe length than the new item, so it will never be swapped anyway. Start hashing with the second item on the collision chain onwards (if any).
• Performance: While filling a hole after a removal, stop rehashing items once we find one whose ideal bucket would be after the hole -- given the Robin Hood hashing invariants, no subsequent items can possibly hash before or at the hole.
• Performance: Update the maximum probe length after a removal by applying a trivial upper bound: the maximum probe length cannot possibly be longer than the count of items remaining. (This is still a quite loose upper bound, but it's extremely cheap to apply.)

Checklist

• I've read the Contribution Guidelines
• My contributions are licensed under the Swift license.
• I've followed the coding style of the rest of the project.
• I've added tests covering all new code paths my change adds to the project (if appropriate).
• I've added benchmarks covering new functionality (if appropriate).
• I've verified that my change does not break any existing tests or introduce unexplained benchmark regressions.
• I've updated the documentation if necessary.

[lorentey]

Cc @tfpauly, with apologies for the oversight 😅