Fix pre-baseline enablement attribution

[meinali-566]

Summary

• keep pre-baseline enablement patches in the reproducible config stack while marking them ineligible for optimization gain attribution
• classify integrate_patch from phase and ownership metadata instead of crediting every patch to Framework Agent
• omit baseline-enablement entries from the canonical optimization projection so warm replay remains the first measured gain

Test plan

• Run focused warm replay and SBD attribution regressions (4 passed)
• Verify pre-baseline enablement promotion through an async manual regression
• Compile all changed Python files
• Run the complete async test selection in CI (local environment lacks pytest-asyncio)

[github-actions]

CI E2E report — ✅ Succeeded

item	value
result	✅ Succeeded
model	Qwen/Qwen3-0.6B (dense)
resources	1× GPU, TP=1
PR branch	fix/warm-replay-enablement-attribution
commit	49f25eb0c408850e86c921463269672681c702ae
session_id	f5b50df9-edd0-47ff-8ed9-a13bf12bafad
queue → dispatch	0s
run time	153m 5s
total	153m 5s

details