[bhelm]

Summary

• register the cross-owner Recovery trace twin in the differential parity ratchet
• add 20 deterministic trace cases covering defaults, statuses, terms, saturation, and exact ordered trace output
• retire only the two now-proven Recovery trace exemptions

Verification

• 277 Swift cases vs 277 Kotlin cases: 0 differences
• Recovery trace slice: 20/20 exact matches
• Python tooling tests: 108/108
• fresh llvm-cov and JaCoCo verification: 44 differential functions, 0 errors
• stale coverage rejected
• score and trace-line mutants detected independently on both platforms
• independent review: no P0/P1/P2 findings

Known boundaries

• Includes the exact negative half-tie regression from [P2] Recovery trace rounds negative half-ties differently on Swift and Kotlin #38.
• Inputs that would exercise the unresolved signed-zero or huge-finite formatting drift in [P2] Recovery trace rounding diverges for signed zero and large finite values #47 fail closed.
• Production code is unchanged.

Closes #38