feat(ws): per-instance event loop, graceful stop, and recv timeout

[hotwater9]

Problem

The ws.Client uses a module-level loop variable that is shared across
all instances. In multi-bot deployments (multiple Client objects in separate
threads), each thread overwrites the same global, causing:

1. RuntimeError: Task got Future attached to a different loop on reconnect (ws.Client: module-level loop variable prevents multiple instances (multi-bot race condition) #119)
2. Silent death — _receive_message_loop crashes but _select() keeps sleeping,
   making the client appear alive while receiving nothing
3. No way to gracefully stop a running client (feat(ws): add public Client.stop() for graceful shutdown #128)
4. Cannot use inside async frameworks ([Python lark-oapi 1.6.7] ws/client.py module-level event loop causes RuntimeError when first imported inside a running loop #133)

Solution

Issue	Fix
Loop collision	self._loop per instance; each start() creates its own loop
Silent death (TCP black hole)	asyncio.wait_for(conn.recv(), timeout=300)
Silent death (task exception swallowed)	_receive_message_loop sets stop_event on unrecoverable failure
No graceful stop	stop() → call_soon_threadsafe(stop_event.set)
Async framework compat	start_async() uses get_running_loop()
Uninterruptible sleeps	All asyncio.sleep() → wait_for(stop_event.wait(), timeout=...)

Breaking Changes

None. start() signature is unchanged. stop() and start_async() are additive.
Module-level loop is retained (set to the last started instance's loop) for
backward compatibility.

Related

• Closes ws.Client: module-level loop variable prevents multiple instances (multi-bot race condition) #119 (multi-bot loop collision)
• Closes [Python lark-oapi 1.6.7] ws/client.py module-level event loop causes RuntimeError when first imported inside a running loop #133 (module-level loop imported inside running loop)
• Supersedes feat(ws): add public Client.stop() for graceful shutdown #128 (graceful stop — this PR includes equivalent functionality)
• Supersedes feat(ws): add Client.start_async() for use inside a running event loop #139 (start_async — this PR includes equivalent functionality)

Testing

• test_multi_instance.py: independent loops, stop() terminates start(),
  stop() interrupts reconnect, start_async() reuses caller loop, backward compat
• Updated test_websockets_compat.py for per-instance loop

[CLAassistant]

All committers have signed the CLA.