fix: resolve MissingGreenlet crash and add MCP fallback in AI Assistant

Capture user_id/account_id before try block so error handler survives
db.rollback() without triggering lazy loads in async context. Add
retry-without-MCP fallback when Anthropic MCP server returns rate limit
or connection errors.

Fixes PYTHON-FASTAPI-3, PYTHON-FASTAPI-4

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

backend/app/services/assistant_chat_service.py

@@ -189,15 +189,29 @@ async def _call_anthropic_cached(
             }
         ]
 
-    response = await client.beta.messages.create(
-        model=settings.AI_MODEL_ANTHROPIC,
-        max_tokens=max_tokens,
-        system=system_blocks,
-        messages=messages,
-        mcp_servers=mcp_servers,
-        tools=tools,
-        betas=["mcp-client-2025-11-20"],
-    )
+    try:
+        response = await client.beta.messages.create(
+            model=settings.AI_MODEL_ANTHROPIC,
+            max_tokens=max_tokens,
+            system=system_blocks,
+            messages=messages,
+            mcp_servers=mcp_servers,
+            tools=tools,
+            betas=["mcp-client-2025-11-20"],
+        )
+    except anthropic.BadRequestError as e:
+        # MCP server failures (rate limits, connection errors) should not
+        # block the assistant entirely — retry without MCP tools.
+        if "MCP server" in str(e) and mcp_servers is not anthropic.NOT_GIVEN:
+            logger.warning("MCP server error, retrying without MCP: %s", e)
+            response = await client.beta.messages.create(
+                model=settings.AI_MODEL_ANTHROPIC,
+                max_tokens=max_tokens,
+                system=system_blocks,
+                messages=messages,
+            )
+        else:
+            raise
 
     # Extract text from response — MCP responses can have multiple block
     # types (text, mcp_tool_use, mcp_tool_result). We join all text blocks.