fix(js_parser): address CodeRabbit review suggestions for Glimmer template support
This PR was written primarily by Claude Code.
Fixes three critical issues identified in CodeRabbit review of PR #8099:
1. **Template tag detection bug**: Fixed byte slice comparison in lexer that was
checking `b"template>"` but using incorrect offset. Changed to `b"template"`
to properly validate characters after `<template`. Now supports whitespace in
opening tags (e.g., `<template >`, `<template\t>`).
2. **Missing lookahead support**: Added `GLIMMER_TEMPLATE` to `is_nth_at_expression()`
to properly recognize templates in expression contexts like `yield <template>...`,
`await <template>...`, etc.
3. **LSP language ID support**: Added "gjs" and "gts" language identifiers to
`try_from_language_id()` for proper LSP integration.
Also enhanced changeset documentation to clarify Phase 1 implementation treats
templates as opaque tokens without internal parsing/linting.
All changes follow TDD approach with tests written first to verify failures,
then fixes applied, and verification that tests pass.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>