feat: Connection pooling and keep-alive

[Snider]

Summary

Reuse HTTP connections for efficiency when collecting from same domain.

Use Case

Collecting 1000 files from same domain - should reuse connections instead of TCP handshake each time.

Implementation

• HTTP/1.1 keep-alive
• HTTP/2 multiplexing
• Connection pool per domain

Configuration

connections:
  max_per_host: 6        # Max connections per domain
  max_idle: 100          # Max idle connections total
  idle_timeout: 90s      # Close idle connections after
  
http2:
  enabled: true
  max_streams: 100       # Max concurrent streams per conn

Commands

# Adjust pool size
borg collect website https://example.com --max-connections 10

# Disable keep-alive (for debugging)
borg collect website https://example.com --no-keepalive

# Force HTTP/1.1
borg collect website https://example.com --http1

Benefits

• Faster collection (no repeated handshakes)
• Lower server load
• Better rate limit compliance

Acceptance Criteria

• Connection pooling per domain
• HTTP/2 support with multiplexing
• Configurable pool sizes
• Idle connection cleanup
• Connection reuse metrics

[google-labs-jules]

Jules is on it. When finished, you will see another comment and be able to review a PR.

[google-labs-jules]

Ready for a review! A PR has been created.