test: add server test exclusion list for experimental parsers
Add NEW_PARSERS_UNSUPPORTED dict to document templates with known issues
when using experimental parsers in server tests:
- LFM2: requires special system message marker
- Llama 3.x: builtin tools need custom TOOL_ARG_NAME handling
- Functionary v3.2: python tool allows raw code fallback
- Nemotron v3: tiny model generates invalid parameter structure
- GPT-OSS: tiny model generates unparseable content
- Kimi K2: tiny model generates format that fails to parse
Also in test-chat.cpp:
- Change test name separator from `_` to `:` for easier grep
- Add skip logic for force_disable_thinking scenarios
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
