Fix/azure container signal handling 6611

[tejas-dharani]

Why are these changes needed?

This PR fixes a critical issue where AutoGen's gRPC runtime host was not working properly in Azure Container Apps and other container environments. The stop_when_signal() method would cause applications to exit immediately with "RuntimeError: Host runtime is not started" instead of waiting for proper signals.
The root cause was limitations of Python's asyncio.loop.add_signal_handler() in container environments, particularly when running in Azure Container Apps where signal delivery behaves differently.

Related issue number

Closes #6611

Checks

• I've included any doc changes needed for https://microsoft.github.io/autogen/. See https://github.com/microsoft/autogen/blob/main/CONTRIBUTING.md to build and test documentation locally.
• I've added tests corresponding to the changes introduced in this PR.
• I've made sure all auto checks have passed.

[shreayan98c]

Hi @tejas-dharani, thank you for your contribution and for addressing the issue #6611 with the stop_when_signal() method in containerized environments. I appreciate the detailed explanation of the root cause and the proposed fix.

I tested the changes in your PR by setting the required environment variables (DOCKER_CONTAINER=true / CONTAINER_APP_NAME) to signal that the environment is a Docker container. I also deployed the updated code, ensuring that AutoGen was installed from the branch containing your fix (fix/azure-container-signal-handling-6611). However, the issue persists, and the runtime host still exits prematurely with the same error: RuntimeError: Host runtime is not started.

Here are some additional observations from my testing:

1. The stop_when_signal() method still seems to behave inconsistently in the Azure Container App environment.
2. I verified that the environment variables were correctly set and accessible within the container.
3. The runtime host starts successfully but does not remain active to listen for events.

Could you please review the changes in the PR to ensure that all edge cases in containerized environments are handled? Additionally, if there are any specific steps or configurations I might have missed during testing, please let me know.