fix: harden backlog refine prompt scaffold and mixed-format parsing

[djm81]

Description

This PR hardens specfact backlog refine --write against real-world Copilot output drift and parser edge cases reported after 0.30.3 rollout.

Key fixes:

• Preserve mixed-format description narrative when inline label-style markers (e.g. **Notes**:) appear before later heading sections.
• Stop label-block capture at markdown heading boundaries to avoid swallowing subsequent sections.
• Keep description fallback only for pure label-style outputs; do not drop valid ## Description heading content.
• Strengthen interactive and export Copilot instructions with an explicit ordered scaffold.
• Explicitly require omission of unknown metadata fields (no placeholders like (unspecified) / no info provided / provide area path).
• Update OpenSpec source tracking to support multiple linked issues for this change (#225, #227).

Fixes #227

New Features #N/A

Contract References: No new/changed @icontract API contracts in this patch.

Type of Change

Please check all that apply:

• 🐛 Bug fix (non-breaking change which fixes an issue)
• ✨ New feature (non-breaking change which adds functionality)
• 💥 Breaking change (fix or feature that would cause existing functionality to not work as expected)
• 📚 Documentation update
• 🔒 Contract enforcement (adding/updating @icontract decorators)
• 🧪 Test enhancement (scenario tests, property-based tests)
• 🔧 Refactoring (code improvement without functionality change)

Contract-First Testing Evidence

Required for all changes affecting CLI commands or public APIs:

Contract Validation

• Runtime contracts added/updated (@icontract decorators on public APIs)
• Type checking enforced (@beartype decorators applied)
• CrossHair exploration completed: hatch run contract-test-exploration
• Contract violations reviewed and addressed

Test Execution

• Contract validation: hatch run contract-test-contracts ✅
• Contract exploration: hatch run contract-test-exploration ✅
• Scenario tests: hatch run contract-test-scenarios ✅
• Full test suite: hatch run contract-test-full ✅

Test Quality

• CLI commands tested with typer test client
• Edge cases covered with Hypothesis property tests
• Error handling tested with invalid inputs
• Rich console output verified manually or with snapshots

How Has This Been Tested?

Contract-First Approach: Added parser/prompt regression tests from user-reported failures, captured failing pre-implementation evidence in OpenSpec TDD log, then implemented fixes and re-ran targeted command/backlog/adapter tests.

Manual Testing

• Tested CLI commands manually
• Verified rich console output
• Tested with different input scenarios
• Checked error messages for clarity

Automated Testing

• Contract validation passes
• Property-based tests cover edge cases
• Scenario tests cover user workflows
• All existing tests still pass

Test Environment

• Python version: 3.11
• OS: Linux

Checklist

• My code follows the style guidelines (PEP 8, ruff format, isort)
• I have performed a self-review of my code
• I have added/updated contracts (@icontract, @beartype)
• I have added/updated docstrings (Google style)
• I have made corresponding changes to documentation
• My changes generate no new warnings (basedpyright, ruff, pylint)
• All tests pass locally
• I have added tests that prove my fix/feature works
• Any dependent changes have been merged

Quality Gates Status

• Type checking ✅ (hatch run type-check)
• Linting ✅ (hatch run lint)
• Contract validation ✅ (hatch run contract-test-contracts)
• Contract exploration ✅ (hatch run contract-test-exploration)
• Scenario tests ✅ (hatch run contract-test-scenarios)

Screenshots/Recordings (if applicable)

N/A