Add lock-free atomic sequence manager to TxManager

[AugistineCreates]

This pr closes #469

Implemented a lock‑free atomic sequence manager and integrated it into the transaction broadcasting flow, eliminating per‑account lock contention.

Overview

• Added src/network/sequence_manager.py with AtomicSequenceManager:

  • Uses Windows InterlockedExchangeAdd (via ctypes) for true lock‑free atomic increments.
  • Fallback to a lock‑protected counter on non‑Windows platforms.
  • Supports block allocation (allocate_block) and per‑thread cached blocks (get_thread_local_block).
  • Exposes a module‑level singleton sequence_manager.

• Updated src/network/tx_manager.py:

  • Replaced per‑account lock and counter with a global, lock‑free sequence fetch:

    from .sequence_manager import sequence_manager as _global_seq_mgr
    sequence = _global_seq_mgr.next()

  • Simplified broadcast flow while preserving signing and dispatch logic.

Benefits

• Performance: Removes the serial lock that previously bottlenecked high‑load broadcasting.
• Scalability: Workers can pre‑allocate blocks of sequence numbers, reducing synchronization overhead.
• Maintainability: Centralized sequence management aligns with the existing nonce_tracker pattern.

Testing & Verification

• All existing tests pass (pytest).
• Manual sanity check confirms correct atomic allocation under concurrent access.
• TxManager.broadcast still logs dispatched transactions with correct sequence numbers.

[drips-wave]

@AugistineCreates Great news! 🎉 Based on an automated assessment of this PR, the linked Wave issue(s) no longer count against your application limits.

You can now already apply to more issues while waiting for a review of this PR. Keep up the great work! 🚀

Learn more about application limits