fix(server): missing timeout on all requests.get() calls — potential indefinite hang

Problem

Multiple requests.get() calls throughout server.py have no timeout parameter, which can cause the server to hang indefinitely if the remote server is unresponsive.

Exact locations

All in minimax_mcp/server.py:

1. voice_clone() — requests.get(file) for URL download (line ~175)
2. voice_clone() — requests.get(response_data.get("demo_audio")).content (line ~205)
3. play_audio() — requests.get(input_file_path).content (line ~220)
4. generate_video() — requests.get(download_url) for video download (line ~330)
5. query_video_generation() — requests.get(download_url) (line ~365)
6. text_to_image() — requests.get(image_url) for image download (line ~410)

Impact

• MCP server can hang indefinitely waiting for a response
• No way to recover without killing the process
• Blocks the entire MCP tool pipeline

Proposed solution

Add a module-level constant and apply to all requests.get() calls:

# At module level
DOWNLOAD_TIMEOUT = 120  # seconds

# Apply everywhere:
requests.get(url, timeout=DOWNLOAD_TIMEOUT)

Also consider adding stream=True for large downloads (video) to avoid loading entire file into memory.

Acceptance criteria

• Every requests.get() call has an explicit timeout parameter
• Large file downloads use streaming where appropriate
• Timeout errors are caught and reported to the user

Start in minimax_mcp/server.py and inspect the six requests.get() call sites listed in voice_clone(), play_audio(), generate_video(), query_video_generation(), and text_to_image(). Use the acceptance criteria as the checklist: every call has an explicit timeout, large downloads stream where appropriate, and timeout errors are caught and reported to the user.