Fix TAG lexer rule to allow a parenthesized value suffix

[lukiod]

Fixes #72.

The TAG lexer rule ('@' ~[@ )]+) excluded ) entirely, so a tag literal ending in a parenthesized value (e.g. @id("<uuid>")) had its closing paren misread as the grouping R_PAREN token instead of part of the tag. That truncates the tag text by one character and leaves a stray ), so tagExpressionMatchesPickle never matches the tag against itself.

Fix: the base tag body now also excludes ( (forcing it into a new optional group instead of being silently absorbed), and an optional trailing ('(' ~[)]* ')')? segment captures a parenthesized value verbatim, including its closing paren, as part of the same TAG token. Existing grouping-paren expressions ((@tagA and @tagB), nested parens) are unaffected, since a bare ( not immediately following @tagName still tokenizes as L_PAREN as before.

Verification: reproduced the failure directly against the unpatched grammar first — confirmed it tokenizes @id("uuid") as TAG=@id("uuid" (missing the final )) plus a stray R_PAREN, so the match fails — then confirmed the fix resolves it and that the existing grouping-paren test cases are unaffected.

One disclosure: this repo's own mvn test -Ptest currently reports 0 tests run for GherkinTagFiltererTest (no junit-vintage-engine on the classpath for this JUnit4-style class) — a pre-existing gap unrelated to this change, so I verified the fix and the three new regression tests directly by compiling against the ANTLR-generated parser and running the exact same assertions standalone rather than through the (currently non-functional) Maven test target.