feat: configuration of foreign TX RPC providers does not require on chain consensus

[DSharifi]

based on #2741

[DSharifi]

@claude review

[claude]

Reviewed the full diff. This PR cleanly replaces the consensus-based vote_foreign_chain_policy with a simpler per-node register_foreign_chain_config registration model. The contract derives supported chains as the intersection of all active participants' registrations.

No critical issues found.

Key observations:

• Rolling update safety: Old vote_foreign_chain_policy now also populates supported_foreign_chains_votes, ensuring mixed old/new node deployments work correctly during rollout.
• Migration correctness: v3_8_1_state.rs seeds the new data structure from the existing on-chain policy (all participants get baseline) and overlays pending votes, preserving intent.
• Unanimity logic is sound: get_supported_foreign_chains requires all active participants to support a chain via is_superset, so stale non-participant entries don't affect results.
• Good test coverage across unit, sandbox, and pytest levels.

✅ Approved