fix: DAO penalty consensus and reset flow

[functor-flow]

What was wrong

• Penalty factor selection was wrong: it could return 100 even when pending votes were 20.
• Penalty processing flip-flopped between apply/reset due to alternating iteration logic.
• Penalty tx flow didn’t wait for finalization.
• One lint warning in torus-bridge test (afterEach unused).

What changed

• Rewrote pendingPenalizations to compute factor from the actual pending vote set (executed=false, deleted_at IS NULL)
  and select nth-highest deterministically.
• Restored reset-to-0 logic, but made it correct:
  • reset only if on-chain penalty >0 and active vote support is below threshold.
• Removed odd/even toggle; now apply+reset candidates are computed each cycle and merged once.
• Updated penalizeAgent to use tx tracker and wait for finalized before success.
• Removed unused afterEach import in bridge test.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
apostle-swarm	Ready	Preview, Comment	Feb 16, 2026 3:32pm
prediction-swarm	Ready	Preview, Comment	Feb 16, 2026 3:32pm
torus-bridge	Ready	Preview, Comment	Feb 16, 2026 3:32pm
torus-governance	Ready	Preview, Comment	Feb 16, 2026 3:32pm
torus-portal	Ready	Preview, Comment	Feb 16, 2026 3:32pm
torus-wallet	Ready	Preview, Comment	Feb 16, 2026 3:32pm