feat: expand multi-harness support for 1.6

[paladini]

Summary

• expand deterministic filesystem detection for Gemini CLI, Kiro, Junie, Roo Code, and Devin
• broaden existing Cursor, Claude Code, Windsurf, Cline, Copilot, Continue, OpenCode, and Antigravity surfaces
• normalize vendor-specific agent, skill, hook, MCP, and hygiene configuration while preserving OR semantics across harnesses
• add 42 focused vendor-surface tests and refresh the golden snapshot
• update the multi-harness documentation in English, Portuguese, Spanish, Hindi, and Chinese

Compatibility and release scope

• keeps the existing check IDs, points, levels, and thresholds unchanged
• remains filesystem-only, deterministic, and dependency-free at runtime
• treats shared standards such as .agents/skills as scoreable but vendor-neutral, avoiding inflated detected-tool counts
• intentionally excludes AGENTS.override.md because it is a local override rather than a portable repository signal
• adds a minor changeset so the release pipeline produces 1.6.0; package versions are not edited manually

Validation

• npm run test:coverage: 21 files, 310 tests passed, 96.39% statement coverage
• npm run check:types -w harness-score
• npm run plugins:sync-check
• npm run docs:build
• npm run lint: 0 errors; 18 pre-existing CSS warnings in untouched files
• npm run scan: L4, 108/108, 100%
• git diff --check

Closes #44

[adeonir]

Hey @paladini, do you have an estimate for when this PR might be ready to merge?

I'm planning to collaborate by adding a Codex plugin, and since this PR is expanding the multi-harness support, I would prefer to build on top of it, to avoid any conflicts.