fix: gracefully stop uvicorn threads during shutdown

[mkg0908]

Summary

• retain the Uvicorn server instance for thread-backed API workers
• request thread shutdown through server.should_exit instead of calling process-only terminate() and kill() methods
• add regression coverage for graceful thread shutdown

Tests

• .venv/bin/python -m ruff check src/litserve/server.py tests/unit/test_lit_server.py
• .venv/bin/python -m compileall -q src/litserve/server.py tests/unit/test_lit_server.py
• focused shutdown regression check using a thread-backed Uvicorn server double

Closes #684

[codecov-commenter]

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

❌ Patch coverage is 85.71429% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 85%. Comparing base (a69b635) to head (dd4e6d5).
⚠️ Report is 4 commits behind head on main.
❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@         Coverage Diff         @@
##           main   #687   +/-   ##
===================================
- Coverage    85%    85%   -0%     
===================================
  Files        39     39           
  Lines      3282   3289    +7     
===================================
+ Hits       2778   2782    +4     
- Misses      504    507    +3     

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[bhimrazy]

The setattr/getattr approach works but I think a small subclass makes the intent clearer and the handle typed:

class _UvicornServerThread(threading.Thread):
    def __init__(self, server: _Server, *args, **kwargs):
        super().__init__(*args, **kwargs)
        self.server = server

w = _UvicornServerThread(
    server, target=server.run, args=(response_queue_id, sockets), name=f"LitServer-{response_queue_id}"
)

wdyt ?

cc: @andyland

[racinmat]

Unfortunately this PR doesn't fix the issue. I just tried it and it was hanging, I got this output.

C:\Projects\others\LitServe\.venv\Scripts\python.exe -X pycache_prefix=C:\Users\Matej.Racinsky\AppData\Local\JetBrains\PyCharm2026.1\cpython-cache C:/Users/Matej.Racinsky/AppData/Local/Programs/PyCharm/plugins/python-ce/helpers/pydev/pydevd.py --multiprocess --qt-support=auto --client 127.0.0.1 --port 19612 --file C:\Projects\others\LitServe\litserve_win_shutdown2.py 
import sys; print('Python %s on %s' % (sys.version, sys.platform))
Connected to pydev debugger (build 261.24374.152)
[MAIN] Starting LitServer...
INFO:     Uvicorn running on http://0.0.0.0:8765 (Press CTRL+C to quit)
C:\Projects\others\LitServe\src\litserve\server.py:1495: UserWarning: Windows does not support forking. Using threads api_server_worker_type will be set to 'thread'
  warnings.warn(
2026-06-06 16:26:11,772 - MainProcess[32988] - litserve.server - DEBUG - server.py:1164 - One or more workers are ready to serve requests
Swagger UI is available at http://0.0.0.0:8765/docs
INFO:     Started server process [32988]
INFO:     Waiting for application startup.
INFO:     Application startup complete.
2026-06-06 16:26:19,972 - MainProcess[32988] - litserve.server - INFO - server.py:1527 - KeyboardInterrupt received. Initiating graceful shutdown.
2026-06-06 16:26:19,972 - MainProcess[32988] - litserve.server - INFO - server.py:1190 - Shutting down LitServe...
2026-06-06 16:26:19,973 - MainProcess[32988] - litserve.server - DEBUG - server.py:1194 - KeyboardInterrupt detected - skipping transport cleanup to avoid hanging

and it never finished.