fix: handle MCP ping at bridge level for servers that don't support it

server-filesystem silently drops ping requests, causing MCP Inspector
and other Streamable HTTP clients to timeout on their first connectivity
check. The bridge now responds to ping directly without forwarding to
the child process.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: andyjmorgan

src/DonkeyWork.CodeSandbox.McpServer/Services/StdioBridge.cs

@@ -187,6 +187,16 @@ void Emit(string eventType, string message, Action<McpStartEvent>? configure = n
 
         // Extract method name for logging
         var method = ExtractMethod(jsonRpcRequest);
+
+        // Handle ping at the bridge level - not all MCP servers respond to ping
+        if (method == "ping")
+        {
+            _logger.LogInformation("Handling ping request id={Id} at bridge level", requestId);
+            lock (_stateLock) _lastRequestAt = DateTime.UtcNow;
+            var rawId = ExtractRawId(jsonRpcRequest);
+            return $"{{\"jsonrpc\":\"2.0\",\"result\":{{}},\"id\":{rawId}}}";
+        }
+
         _logger.LogInformation("Sending request id={Id} method={Method}, registering TCS", requestId, method);
 
         // Register the TCS BEFORE writing to stdin to avoid a race where
@@ -638,6 +648,12 @@ private static bool IsNotification(string json)
         }
     }
 
+    private static string ExtractRawId(string json)
+    {
+        using var doc = JsonDocument.Parse(json);
+        return doc.RootElement.GetProperty("id").GetRawText();
+    }
+
     private static string? ExtractMethod(string json)
     {
         try