Regression on value class parameter handling

[sgerke-1L]

The fix in #36449 introduces a regression related to the handling of value class conversion. In particular, passing e.g. UUID for an argument of type ValueClass(val value: UUID) doesn't work anymore. This is required to make ValueClass parameters work, see added integration tests.

Fix description

• Partially reverts Handle Kotlin nullable value class param correctly in CoroutineUtils #36449 to fix the aforementioned regression
• Fixes the issue reported in Handle Kotlin nullable value class param correctly in CoroutineUtils #36449 by introducing a !JvmClassMappingKt.getJavaClass(kClass).isInstance(arg) guard in the appropriate places, and keeps the regression test in from Handle Kotlin nullable value class param correctly in CoroutineUtils #36449
• Introduced integration tests that test the handling of Kotlin value classes and the interaction of optional request parameters with an optional value class argument

[sdeleuze]

I will try to leverage this PR to fix this regression caused by #36449, the fix will have to be backported to 6.2.x.

[sgerke-1L]

Thank you! Let me know if there is anything I can help with.

[sdeleuze]

Were the integration tests required to exercise some broken code paths or were they just added as additional tests "in case of" ?

[sgerke-1L]

Suspending handler method with nullable value class request param and suspendingValueClass exhibit the regression introduced in #36665 . The others were green already, and I added them to document what's currently working.