fix(serve): honor container_timeout_seconds in LocalContainerMode.create_server

[AkshayShah03]

Issue #, if available: N/A

Description of changes:

LocalContainerMode.create_server overwrote its container_timeout_seconds argument before computing the health-check deadline:

log_generator = self.container.logs(follow=True, stream=True)
container_timeout_seconds = 1200
time_limit = datetime.now() + timedelta(seconds=container_timeout_seconds)

ModelBuilder.deploy(container_timeout_in_seconds=N) in LOCAL_CONTAINER mode passes N down to create_server, but the value was ignored:

• a short timeout still blocked for 20 minutes before failing;
• a model that needs more than 20 minutes to start couldn't be given more time, even though the failure message says "Please increase container_timeout_seconds".

This change removes the override, so the requested timeout is used. Callers that don't pass one now get the documented default of 300 seconds, the same as v2, instead of the hard-coded 1200.

Testing:

• New tests/unit/mode/test_local_container_mode_timeout.py runs create_server with a fake clock and an always-unhealthy ping. It checks the wait ends at the requested timeout, for both 60 s and 1800 s.
• Before this change both cases stopped at 1220 s.
• tests/unit/mode: 12 passed.

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.