Fix stale size in AbstractMapMultiSet view iterator remove

[rootvector2]

AbstractMapMultiSet's uniqueSet() and entrySet() view iterators drop the whole entry from the backing map but never decrement the cached size field, so removing an element with N copies through either view leaves size() == N while isEmpty() is true and toArray() (sized by size()) returns N trailing nulls.

UniqueSetIterator.remove() calls super.remove() (which deletes the key) and then parent.remove(lastElement, count), but that call short-circuits at map.get(...) == null and never adjusts size; EntrySetIterator.remove() calls decorated.remove() and never touches size at all. Both overrides now subtract the removed entry count from size and bump modCount, matching the size-correct AbstractMapBag.BagIterator and remove(Object, int). Found by comparing the two map-backed count collections. HashMultiSet and TreeMultiSet inherit the fix.

Repro before the patch:

MultiSet<String> ms = new HashMultiSet<>();
ms.add("a", 3);
Iterator<String> it = ms.uniqueSet().iterator();
it.next();
it.remove();
ms.size();    // 3  (expected 0)
ms.toArray(); // [null, null, null]

• Read the contribution guidelines for this project.
• Read the ASF Generative Tooling Guidance if you use Artificial Intelligence (AI).
• I used AI to create any part of, or all of, this pull request. Which AI tool was used to create this pull request, and to what extent did it contribute?
• Run a successful build using the default Maven goal with mvn; that's mvn on the command line by itself.
• Write unit tests that match behavioral changes, where the tests fail if the changes to the runtime are not applied. This may not always be possible, but it is a best practice.
• Write a pull request description that is detailed enough to understand what the pull request does, how, and why.
• Each commit in the pull request should have a meaningful subject line and body. Note that a maintainer may squash commits during the merge process.

[garydgregory]

Hello @rootvector2

• Please read the MultiSet contract and it seems like this PR would break it. WDYT?
• This PR changes a main abstract class but only tests a single implementation. If this PR is an actual fix (see above), we need to test all concrete subclasses.
  TY!

[rootvector2]

The removal behavior is unchanged and still follows the contract. uniqueSet() and entrySet() removal drops the whole entry (all occurrences of the element), exactly as before. The patch only fixes the cached size field, which those two view iterators never decremented. That left size() reporting N while isEmpty() returned true and toArray() (sized by size()) came back with N trailing nulls, so size() was inconsistent with the other views. Found it comparing this class against AbstractMapBag, whose bag iterator already adjusts size.

On testing: moved the regression test up into AbstractMultiSetTest so it runs for every concrete implementation (HashMultiSet, PredicatedMultiSet, SynchronizedMultiSet; skipped for UnmodifiableMultiSet, which does not support removal) instead of just HashMultiSet.

[garydgregory]

Thank you @rootvector2 , merged 🚀 Note that I added 2 assertions to check that remove() can only be called once per next().