feat(skillopt): add meta-skill module for optimizer cross-epoch memory

[osilkin98]

Closes #1236

Changes

• New skillopt/ package — adds optimizer-side meta-skill memory for cross-epoch learning
• Core module (skillopt/optimizer/meta_skill.py): format_meta_skill_context() renders optimizer memory into prompt-ready blocks; run_meta_skill() calls the optimizer model with adjacent-epoch comparison data; load_meta_skill_content() loads previous epoch results with 3-epoch recency window; validate_deployment_gate() catches meta-skill leakage in final SKILL.md; should_generate_meta_skill() enforces epoch-2 minimum and score-delta conditioning
• Prompt template (skillopt/prompts/meta_skill.md): optimizer-coach system prompt that produces abstract optimizer-facing guidance, not target-facing task instructions
• Trainer hooks (skillopt/engine/trainer.py): generate_epoch_meta_skill() epoch-boundary hook with resume safety, first-epoch skip, and score-delta conditioning; load_active_meta_skill() epoch-start loading
• Reflect injection (skillopt/gradient/reflect.py): reflect_on_errors(), reflect_on_successes(), reflect_and_merge() — all accept meta_skill_context parameter for in-memory prepend to optimizer prompts (on-disk analyst templates are never modified)
• Safety constraints: 3000 token cap, 3-epoch recency window, score-delta conditioning, deployment gate validation, file separation (meta_skill/ directory, never skills/)
• 49 unit tests covering all key functions including edge cases (malformed JSON, token overflow, recency boundary, negative delta skip)

Mirrors the architecture from the reference implementation at ~/SkillOpt/ but adapted for the factory's module structure and dependency injection pattern (explicit chat_fn parameter instead of global import).

[github-actions]

Sentrux Quality Report

Absolute

Scanning ....
[scan] git ls-files: 619 total, 607 kept, 12 dropped (ext:12, meta:0, big:0)
[build_project_map] 607 files, 107 unique dirs, 99 cache misses, 6.1ms
[resolve] 1097 resolved, 1546 unresolved (of 2643 total specs)
[resolve_imports] project_map 6.2ms, suffix_idx 1.1ms, suffix_resolve 17.7ms, total 25.0ms
[build_graphs] 607 files | maps 2.2ms, imports 25.2ms, calls+inherit 5.9ms, total 33.3ms | 1096 import, 8682 call, 11 inherit edges
sentrux check — 3 rules checked

Quality: 4692

✗ [Error] max_cc: 3 function(s) exceed max cyclomatic complexity of 30
    factory/cli/_ceo_helpers.py:_validate_ceo_flags (cc=43)
    factory/cli/_ceo_helpers.py:_execute_ceo (cc=43)
    factory/cli/run.py:cmd_run (cc=32)

✗ 1 violation(s) found

Diff (vs base branch)

Scanning ....
[scan] git ls-files: 619 total, 607 kept, 12 dropped (ext:12, meta:0, big:0)
[build_project_map] 607 files, 107 unique dirs, 99 cache misses, 6.4ms
[resolve] 1097 resolved, 1546 unresolved (of 2643 total specs)
[resolve_imports] project_map 6.5ms, suffix_idx 1.1ms, suffix_resolve 17.6ms, total 25.3ms
[build_graphs] 607 files | maps 2.2ms, imports 25.4ms, calls+inherit 6.6ms, total 34.2ms | 1096 import, 8682 call, 11 inherit edges
sentrux gate — structural regression check

Quality:      4694 -> 4692
Coupling:     0.79 → 0.79
Cycles:       3 → 3
God files:    3 → 3

Distance from Main Sequence: 0.38

✓ No degradation detected

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 86.24%. Comparing base (1925cdf) to head (e051c72).
⚠️ Report is 138 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1271      +/-   ##
==========================================
- Coverage   88.22%   86.24%   -1.99%     
==========================================
  Files         139      211      +72     
  Lines       15524    23096    +7572     
  Branches     2433     3640    +1207     
==========================================
+ Hits        13696    19919    +6223     
- Misses       1308     2333    +1025     
- Partials      520      844     +324     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.