fix: include token type in unexpected token error messages

[Pastalikek65]

Unexpected token errors now report the symbolic name of the lexer's token type when one exists, e.g.:

unexpected token "group" of type Keyword (expected <ident>)

Token types without a symbolic name (such as literals) keep the previous format, e.g.:

unexpected token ")" (expected <ident>)

This is the format suggested by @alecthomas in #265 ("unexpected token "group" of type keyword").

Implementation notes:

• parseContext now carries the lexer's symbol table (built once from the parser's lexer.Definition); every UnexpectedTokenError construction point (7 sites across parser.go/nodes.go) fills the new TokenType field.
• UnexpectedTokenError.TokenType is ordered after the pre-existing exported fields, so external positional struct literals remain source-compatible.

Verification: go test ./... (root + _examples), golangci-lint run — no new issues vs. baseline.

Fixes #265.

[alecthomas]

Mmm. I think this should use the same <lower-symbol> syntax that we already use elsewhere for grammars.

[Pastalikek65]

Thanks @alecthomas! I've updated the format to use the same <lower-symbol> syntax as elsewhere in the grammar (e.g. unexpected token "struct" of type <ident>), via tokenTypeName returning <lower>-wrapped symbols. All tests updated and passing.