Make WrapDynaClass instance cache thread-safe

[rootvector2]

WrapDynaClass.createDynaClass populates its per-classloader cache with computeIfAbsent, but the cache is a plain WeakHashMap, so threads that share a context classloader race on an unsynchronized map where computeIfAbsent is not atomic: concurrent callers can build and return different WrapDynaClass instances for one key, and the structural writes can corrupt the map. Switch the cache to BeanUtils.createCache(), the weak-keyed ConcurrentReferenceHashMap that MethodUtils, ConvertUtilsBean and PropertyUtilsBean already use. Found while auditing the package's computeIfAbsent call sites against those thread-safe caches. Like the MethodUtils visibility fix this race has no deterministic unit test so none is added, though a stress loop over concurrent createDynaClass calls does surface the duplicate instances.

• 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]

@rootvector2
Please add a unit test to avoid a future regression.

[rootvector2]

added a regression test that races 16 threads on createDynaClass for one bean class over 200 rounds and asserts a single cached instance comes back. it fails against the old WeakHashMap and passes with BeanUtils.createCache().

[garydgregory]

Merged, TY @rootvector2 , please port to 1.X if possible.

[rootvector2]

thanks for merging. port is up in #419 — 1.X has no atomic computeIfAbsent to lean on, so it swaps the cache to BeanUtils.createCache() and synchronizes the get/create/put in createDynaClass, with the same regression test ported.