RubyParser: emit ASSIGNMENT_LIKE_METHOD_IDENTIFIERS

[xavierpinho]

Previously, so-called assignment-like method identifiers (e.g. foo2= in def foo2=(x) ... end) were being lexed as two separate tokens (LOCAL_VARIABLE_IDENTIFIER and EQ) and handled together in a parser rule. However, with the introduction of so-called endless methods, this separation could lead to wrong parse trees, as witnessed by #3270 .

To resolve this ambiguity, we shall now gobble up the EQ token as part of the method name, emitting an ASSIGNMENT_LIKE_METHOD_IDENTIFIER token instead.

Closes #3270