feat(voteballots): add quick answer generation from confidence-scored data (#713)

* feat(voteballots): add quick answer generation from confidence-scored data

WSP 97 constraints enforced:
- NO_LLM_CALL: Pure template-based generation
- NO_NEW_FACTS: Only surfaces existing labeled data
- MAX_3_LINES_ENFORCED: Truncates with human review note

Adds quick_answer.py module with:
- QuickAnswer dataclass with provenance tracking
- generate_quick_answer() for template-based answer generation
- format_funding_line() for confidence-annotated source lines
- truncate_with_review_note() for MAX 3 line enforcement
- Shell, markdown, and plain text output formats

Test coverage: 46 new tests (241 total voteballots tests)
- Verified fact clean answers
- Low confidence uncertainty markers
- Truncation enforcement
- Human review flag preservation
- Trail termination handling
- No LLM call contract verification
- Safety boundary tests (no recommendations, no persuasion)

Slice 5 of Vote PoC Chain Lane.

Worker-Lane: W6
Slice: VOTE_POC_QUICK_ANSWER_GENERATION_PHASE1

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

* fix(voteballots): add missing Slice 5 artifacts for PR #713

Repair for W10 gate block:
- src/__init__.py: Export Slice 5 carry-forward contract (6 items)
- ModLog.md: Append WSP 22-compliant entry for Slice 5
- Audit doc: VOTE_POC_QUICK_ANSWER_GENERATION_PHASE1.md with
  * Dependencies: #707, #709, #710, #712 (all merged)
  * Internal Review Verdict: READY
  * Truth Boundary Checklist: 30/30 YES
  * Disallowed language scan: PASS
  * Carry-forward contract for Slice 6

All Slice 1-4 exports preserved. 241/241 tests pass.

Worker-Lane: W6
Slice: VOTE_POC_QUICK_ANSWER_GENERATION_PHASE1

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

---------

Co-authored-by: Foundups Agent <dev@foundups.com>
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>

Author: 3 people