Clean up Python cache/build files from atlassian-mcp-server

## Issue Description

The forked atlassian-mcp-server has several files that shouldn't be in source control. Since we're no longer packaging this MCP server for external distribution, we should clean up these files.

## Files to Clean Up

### 1. Virtual Environment
The entire `.venv/` directory is tracked in git:
- `mcp/servers/atlassian-mcp-server/.venv/` (thousands of files)

### 2. Python Cache Files
78 `.pyc` files in the `src/` directory:
- `mcp/servers/atlassian-mcp-server/src/mcp_atlassian/__pycache__/*.pyc`
- `mcp/servers/atlassian-mcp-server/src/mcp_atlassian/*/__pycache__/*.pyc`

### 3. Modified Cache Files
Currently showing as modified:
- `mcp/servers/atlassian-mcp-server/src/mcp_atlassian/jira/__pycache__/client.cpython-312.pyc`
- `mcp/servers/atlassian-mcp-server/src/mcp_atlassian/jira/__pycache__/projects.cpython-312.pyc`

## Proposed Solution

1. **Remove all tracked cache/build files**:
```bash
git rm -r --cached mcp/servers/atlassian-mcp-server/.venv/
git rm -r --cached mcp/servers/atlassian-mcp-server/src/**/__pycache__/
```

2. **Verify .gitignore is properly configured** (already has proper entries for `.venv/`, `__pycache__/`, `*.pyc`)

3. **Consider if any other build/dist files need cleanup**

## Questions to Consider

- Should we keep the .gitignore file as-is for potential future packaging?
- Are there any other distribution-related files (setup.py, pyproject.toml configs) that could be simplified?

## Context

This cleanup is needed because the MCP server was forked and committed with these files already tracked. The existing `.gitignore` has the right patterns, but git continues tracking files that were already committed.

Start with the tracked paths under mcp/servers/atlassian-mcp-server/.venv/ and src/mcp_atlassian/**/__pycache__/, then inspect the existing .gitignore entries for .venv/, __pycache__/, and *.pyc. Remove the tracked cache and build files, check for other distribution-related files, and verify that the cleanup leaves the ignore rules intact.

Written by the indexing model from the issue text.