fix: allow a final modifier on a switch pattern's type pattern

[DragonAxe]

instanceof_expression already accepts optional('final') before its pattern's type, matching JLS support for final on a pattern variable in both instanceof patterns and switch patterns (JEP 394/405, JEP 441/440). type_pattern -- the equivalent rule used inside a switch's case labels -- never got the same treatment, so case final Type x -> produced an ERROR node: the parser had no rule for a leading final keyword there, so it consumed final itself as if it were the pattern's type name and the real type as the binding identifier, leaving the actual variable name as a stray ERROR token.

case Type x -> (no modifier) and instanceof final Type x both already parsed correctly; this just brings case final Type x -> in line with them. All existing grammar tests still pass.

Fixes #229

[DragonAxe]

Disclaimer: This fix was found and documented by Claude while I was working on an auto-formatter for java. I apologize for any inconvenience this might cause. I, the author, am willing to make any adjustments or clarifications if needed.

[wetneb]

Hi @DragonAxe,
Since external pull requests are rarely integrated in this repository, we have formed a small group of contributors and users to maintain a fork of this grammar. I looked into integrating your change to our fork, but it turns out that this issue is already solved there.

If you encounter any issue switching to our fork as a user or contributor, I would be happy to hear what we could improve to enable that. Since you seem to be using this grammar via the Rust crate, you should hopefully be able to just switch to https://crates.io/crates/tree-sitter-java-orchard.