fix(transaction): prevent concurrent map access in monitor

[mfw78]

Checklist

• I have read the coding guide.
• My change requires a documentation update, and I have done it.
• I have added tests to cover my changes.
• I have filled out the description and linked the related issues.

Description

Fixes a race condition in pkg/transaction/monitor.go that causes fatal error: concurrent map iteration and map write panics.

The checkPending() function was iterating over watchesByNonce without holding the lock, while WatchTransaction() could concurrently modify the map. Classic Go map concurrency footgun. 🔫

Solution: Snapshot-based approach that minimizes lock contention:

1. Phase 1 (lock held): Snapshot transaction data needed for RPC calls
2. Phase 2 (lock released): Make slow RPC calls (TransactionReceipt, NonceAt)
3. Phase 3 (lock held): Notify subscribers and cleanup

This avoids blocking WatchTransaction() callers during potentially slow (100-500ms) RPC operations.

Open API Spec Version Changes (if applicable)

N/A

Motivation and Context

Discovered while running Bee nodes in a Kurtosis test environment. The panic occurred during chequebook deployment when transaction monitoring was active.

fatal error: concurrent map iteration and map write

goroutine 182 [running]:
github.com/ethersphere/bee/v2/pkg/transaction.(*transactionMonitor).checkPending(0x4000ab8d00, 0xd46)
        /app/pkg/transaction/monitor.go:198 +0x334

Related Issue (Optional)

N/A

Screenshots (if appropriate):

N/A