Add genesis migration dual-write queue for new Core chain

[rickyrombo]

Summary

• After a relay transaction is confirmed, enqueue it (fire-and-forget) into new_chain_queue
• NewChainFlusher goroutine drains the queue and forwards transactions to the new Core chain via ForwardTransaction ConnectRPC

Ordering

Transactions are flushed sequentially in insertion order. Parallel flushing was considered and rejected: dev apps can act on behalf of other users, creating cross-user ordering dependencies that no simple partitioning scheme can safely handle.

On error the flusher halts at the failing row (2s backoff + retry) rather than skipping — skipping would violate ordering. Mempool-full responses are retried separately with exponential backoff (500ms → 15s cap).

Config knobs (all off by default)

Env var	Purpose
newChainUrl	ConnectRPC endpoint of the bootstrap node
newChainQueueEnabled	Turns on enqueueing in the relay path
newChainFlushEnabled	Starts the background flusher goroutine
newChainFlushFromBlock	Trims rows already covered by the genesis backfill
newChainInsecureSkipVerify	Disables TLS verification (for localstack / self-signed certs)

Test plan

• Unit tests cover: enqueue correctness, backfill trim, trim+flush, full drain
• Deploy with all knobs off — no behaviour change
• Enable newChainQueueEnabled first to populate the queue before enabling the flusher

🤖 Generated with Claude Code