Retrieve Metadata tool is going into race condition

> [!Important]
> **Before you submit your issue, make sure that:**
> - You're using the latest version of Salesforce DX MCP Server.
> - This can be force by using `@salesforce/mcp@latest` in your MCP config json
> - You've searched both open and closed issues for related posts.
> - You understand that GitHub Issues don't adhere to any agreement or SLA.
> - If you require immediate assistance, use official channels such as Salesforce Customer

### Summary

We have couple of developers who are trying to do the retrieve metadata from org using the mcp tool it is going into race condition and executing as batch. They are getting from one org concurrently.
With these they are waiting more than 30 mins to 1 hour which is impacting the work.
### Steps To Reproduce
Teams will prompt to get metadata from one org concurrently.
The tool goes into batch mode and it waits till timeout and retires back.

### Expected Results

We need everyone to get it in less time and proper results

### Actual result

### Additional information

### Environment details
- Salesforce DX MCP Server version (run `npx --offline @salesforce/mcp --version`)
- v0.30.9
- MCP Client (A4D, Copilot, Cursor, etc)
- Windsurf
- LLM Model used (GPT-4.1, Claude Sonnet 4, ect)
- Claude / GPT

No source file or test is named. Start by reproducing concurrent metadata retrieval from one org with Salesforce DX MCP Server v0.30.9 and Windsurf, using the reported version command to verify the environment. Done means concurrent requests return proper results without entering batch mode, waiting for the timeout, or retrying.

Written by the indexing model from the issue text.