Improve exception handling for ClientDisconnect errors

[FanisPapakonstantinou]

Fix: Handle ClientDisconnect gracefully (return 499, not 500)

Problem

ClientDisconnect exceptions are currently handled incorrectly in StreamableHTTPServerTransport._handle_post_request:

• Returns HTTP 500 (Internal Server Error)
• Logs as ERROR with full traceback
• Triggers production 5XX alerts

ClientDisconnect occurs during normal operations when clients disconnect mid-request due to:

• Network timeouts
• User cancellations
• Load balancer timeouts
• Mobile network interruptions

These are client-side events, not server failures, and should not be treated as server errors.

Root Cause

The broad except Exception handler in _handle_post_request catches ClientDisconnect before it can be handled appropriately, resulting in incorrect error logging and HTTP status codes.

File: src/mcp/server/streamable_http.py (lines ~490-500)

Solution

This PR adds a specific ClientDisconnect exception handler before the broad Exception handler:

1. Catches ClientDisconnect explicitly
2. Returns HTTP 499 (Client Closed Request) instead of 500
3. Logs as INFO instead of ERROR
4. Follows the same pattern as other 4xx responses in the file