fix: detect sponsored tx nonce conflicts using the sponsor nonce

[rafa-stacks]

Replace-by-fee detection and mempool prune/restore keyed sponsored transactions by (sponsor_address, nonce). But nonce is the origin's nonce, and sponsor_nonce was never read. A sponsored tx actually uses two account nonces: (origin, nonce) and (sponsor, sponsor_nonce). The node's mempool enforces a unique constraint on each.

This PR computes conflicts per nonce slot. Each tx occupies (sender, nonce), plus (sponsor, sponsor_nonce) when it's sponsored.

[codecov]

Codecov Report

❌ Patch coverage is 99.24242% with 1 line in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
migrations/1779800000026_sponsor-nonce-indexes.ts	97.22%	0 Missing and 1 partial ⚠️

📢 Thoughts on this report? Let us know!

[Copilot]

Copilot review overview

🟡 Changes recommended

Independent winner selection can incorrectly prune valid transactions when sponsored nonce-slot conflicts overlap.

Review effort: Balanced
Findings: 1

Open (1)

• Pruned historical transactions can incorrectly win mempool slots · New

What changed in this PR

Corrects sponsored transaction conflict detection to use both origin and sponsor nonce slots.

Changes:

• Reworks RBF, pruning, and restoration around nonce slots.
• Adds sponsor-nonce indexes.
• Adds sponsored transaction conflict and reorg tests.

File	Description
src/​datastore/​pg-write-store.ts	Implements slot-based conflict handling.
migrations/​1779800000026_sponsor-nonce-indexes.ts	Replaces incorrect sponsor indexes.
tests/​api/​mempool/​mempool.test.ts	Covers sponsored nonce conflicts and reorgs.
tests/​api/​test-builders.ts	Supports sponsor nonces in fixtures.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[Copilot]

Copilot review overview

🟡 Changes recommended

Winner selection and re-org restoration can still leave the mempool in an incorrect state.

Review effort: Balanced
Findings: 1

Open (1)

• Compute consistent winners before pruning conflicting transactions · New

Resolved since last review (1)

• Pruned historical transactions can incorrectly win mempool slots

Previously missed (1)

In code that hasn't changed since last review

Recalculate RBF ordering for restored mempool transactions

src/​datastore/​pg-write-store.ts:4283

When every orphaned transaction already has a mempool row (for example, a transaction that was pending before confirmation), this CTE restores all same-slot rows to Pending but never recalculates RBF ordering. The later block-ingestion call is keyed only by transactions in the replacement block, so an empty replacement block leaves all conflicting rows visible and overcounts mempool_tx_count. The current re-org tests avoid this by confirming a transaction that never touched the mempool, which forces the insertion path to run the recalculation. Re-run RBF for the restored IDs here as well.