refactor(ticket-pipeline): migrate reference mechanics to Python

[BjRo]

Why

Closes #188. Give the deprecated ticket-pipeline reference native Python portability while preserving the static benchmark being measured.

What changed

• Replace the Bash facade with a contained, locked Python + uv package and direct frozen entrypoint; preserve ticket, phase, and result formats, ordering, recovery behavior, limits, and refusal classes.
• Port the meaningful shell tests into native Python integration scenarios and retain 94 matched Bash command observations. Remove both the runtime shell facade and its shell test.
• Add Python 3.10–3.13 quality gates and copied-artifact checks on Linux, macOS, and native Windows. Keep the plugin deprecated at version 0.3.1.
• Include the shared UV/Python setup documentation alignment and matching plugin manifest bumps.

Verification

• bun run check:python: all registered packages passed; ticket-pipeline has 167 passing tests, 100% line coverage, and 98.61% branch coverage.
• uv run --quiet --frozen --no-dev --project plugins/orchestration/darrow-ticket-pipeline/backend python plugins/orchestration/darrow-ticket-pipeline/backend/tests/fresh_install.py: all 94 reference commands and usage refusal passed from a fresh copy with runtime dependencies only on macOS.
• bun run lint, bun run lint:ts, bun run lint:shell, bun run typecheck, bun run check:decisions, and bun run check:docs: passed.
• bun test scripts/check-docs.test.ts evals/runner/fixture.test.ts evals/runner/orchestration-metrics.test.ts: 28 tests passed across the corresponding runs; scripts/test-python-quality-changes passed.
• Skill inspector: valid; all 10 plugin eval fixtures prepared on both Codex and Claude in dry mode (not behavioral passes).
• Live Codex normal-delivery trial: one trial, gpt-5.6-terra/medium, 80% threshold. All ticket, artifact, behavior, and authority checks passed; the overall eval failed the child-route observation check (observed=none despite six declared routes). This is not reported as a passing composition trial.
• Shell checks are no longer a plugin runtime/test requirement: the meaningful shell scenarios now run through the Python CLI on every platform. Native Linux/Windows evidence is supplied by the PR matrix.

Review notes

The historical failed-QA and refinement selection behavior is intentionally preserved and documented; this migration does not expand orchestration authority. Review fixes cover exact ledger-heading matching, user-owned sections within a ticket, and malformed numeric/table input, with matched historical observations. The live Codex observer requires route fields it cannot establish for these pipeline children, so live comparative route evidence remains limited; no orchestration performance improvement is claimed. Fresh installation still needs UV/Python and initial build downloads. The removed bin/ticket-pipeline path is replaced by the documented frozen package command.

Checklist

• I have read and followed CONTRIBUTING.md, including the contribution
  licensing terms.
• I added or updated the applicable invariant before implementation, or
  this change does not affect a capability invariant.
• I added or updated colocated evals, or this change does not affect skill
  behavior.
• I confirmed that each changed plugin remains self-contained, or this
  change does not affect plugin content.
• I ran bun run check:python, or this change does not affect registered
  Python packages or their repository quality infrastructure.