[java] AST inconsistencies around FormalParameter

[oowekyala]

The following are breaking API changes, that I'd like to see done sometime

FormalParameter shouldn't be an access node

FormalParameter currently implements AccessNode, even though this is only because a parameter can be final...

The problem is also real for LocalVariableDeclaration

CatchStatement shouldn't use FormalParameter as a child

• Done in [java] Extract CatchParameter from FormalParameter #2202

Because of that, FormalParameter's grammar implements the multicatch syntax, which doesn't really make sense and makes it difficult to enrich the API of the node in a sensible way.

• For example, methods that are only relevant to the formal parameter usage make no sense if the node is a catch statement (e.g. isVarargs), and conversely (isMultiCatch).
• For methods that can be used in both cases, the documentation is very complicated to write
• Most importantly, type resolution is inconsistent between the two cases

We should probably follow the JLS more closely for Catch formal parameters.

Resource shouldn't extend FormalParameter

• Done in [java] Improve try-with-resources grammar #1897

It just doesn't make sense at all. ASTResource inherits isVarargs, isMultiCatch, all methods of AccessNode, etc...

[oowekyala]

The last two points are fixed on the java-grammar branch, the first point is a subset of #1307, so I'll close this