Clean up distributed app builders to fix test failures

[JamesNK]

Problem:

1. Test creates DistributedApplicationBuilder.
2. DistributedApplicationBuilder creates HostApplicationBuilder.
3. HostApplicationBuilder creates configuration, including a FileSystemWatcher.
4. Test never builds builder and never disposes of host. The FileSystemWatcher hangs around until finalizer runs.

There are many tests like this. Too many watchers exceeds a Linux limit and errors start.

Fix proposed here:

• Have an internal method on DistributedApplicationBuilder to clean up HostApplicationBuilder's config. Tests will need to be updated to use it.
• Have a test helper type BuilderContainer that makes it easy to create and "dispose" builders.
• Have TestDistributedApplicationBuilder that wraps an inner DistributedApplicationBuilder. It implements dispose, and disposing the test builder will create a host and then dispose it.

Only tests in AWSCloudFormationResourceTests are updated. All tests that create a builder but never create a host that is disposed will need to be changed.
I updated the tests I could find that didn't build and dispose a host.

Microsoft Reviewers: Open in CodeFlow

[radical]

Suggested change

	public sealed class TestDistrubtedApplicationBuilder : IDisposable, IDistributedApplicationBuilder
	public sealed class TestDistributedApplicationBuilder : IDisposable, IDistributedApplicationBuilder

[joperezr]

Thanks for fixing this James!