Performance: LDAP and SMB connection pooling

Description

Implement connection pooling for LDAP and SMB connections to avoid repeated connection overhead during parallel tool execution.

Current Behavior

• Each tool creates new LDAP connection
• Each tool creates new SMB connection
• Redundant TLS handshakes
• Redundant NTLM negotiation

Proposed Implementation

1. LDAP Connection Pool

   • Reuse anonymous bind connections across tools
   • RootDSE queries share connection
   • Per-DC pool (multiple DCs may be discovered)
   • Connection timeout and re-establishment

2. SMB Connection Pool

   • Reuse authenticated SMB sessions
   • Share enumeration reuses connections
   • Null session pooling (if available)
   • Automatic reconnection on timeout

3. Pool Management

   • Max 10 connections per DC (configurable)
   • Idle timeout: 300 seconds
   • LRU eviction policy
   • Health check before reuse

Implementation

• Create: adpentest/connection_pool.py
• Singleton pattern for pool management
• Thread-safe with locks
• Metrics tracking (hit rate, misses, timeouts)

Files to Update

• adpentest/core.py - Use pooled connections
• Create: adpentest/connection_pool.py

Success Criteria

• 50%+ reduction in connection overhead
• No connection pool exhaustion under 16 concurrent tools
• Transparent to existing code

Start by reading adpentest/core.py to trace how LDAP and SMB connections are created during parallel tool execution. Then design the thread-safe singleton in the new adpentest/connection_pool.py, including per-DC pooling, limits, timeouts, reconnection, eviction, health checks, and metrics. Done means core.py uses the pools transparently and the stated concurrency and connection-overhead criteria are met.