wrap-java: Supporting return type erasure bounds

[sidepelican]

This PR is part of the improvements for #599.

Calling getGenericDeclaration() on a generated java.lang.reflect.TypeVariable currently results in a runtime crash:

JavaLangReflect/TypeVariable.swift:14: Fatal error: Java call threw unhandled exception: java.lang.NoSuchMethodError: Lsun/reflect/generics/reflectiveObjects/TypeVariableImpl;.getGenericDeclaration()Ljava/lang/Object;

This is because the JNI method lookup is searching for a signature returning Object, whereas the actual erased signature in Java returns GenericDeclaration (the upper bound of the type parameter D).

Currently, the @JavaMethod macro expands using JavaObject?.self as the resultType for the erased generic return type.
Since the macro cannot infer the original Java bounds from the Swift method signature alone, it defaults to Object.

I have introduced a new macro argument, typeErasedResultBound, to explicitly provide the bound type.

With this change, we can specify the correct type for JNI method resolution:

@JavaMethod(typeErasedResult: "D!", typeErasedResultBound: GenericDeclaration?.self)
public func getGenericDeclaration() -> D!

  let result$ = {
    do {
      return try dynamicJavaMethodCall(methodName: "getGenericDeclaration", resultType: /*type-erased:D*/ GenericDeclaration?.self)
    } catch {
  ...

[ktoso]

Could need some docs why this one does require the erased result types and then do a See: to the "main one" for more docs.

[ktoso]

Actually... how about we keep ONE macro, and diagnose from inside the macro if typeErasedResultBound was provided but the typeErasedResult wasn't? Since that's what you're trying to enforce here. I'd like to avoid having many overloads here in the sources so that docs are easy to find

[ktoso]

Very nice, one nitpick about the new macro declaration, best if we could keep the single declaration, wdyt?

[sidepelican]

I was able to unify the macro definitions.
I have added an explanation for typeErasedResultBound. Let me know what you think.

[ktoso]

Very nice, LGTM; only nitpickt that the docs were not showing a wildcard, a wildcard is ?