fix: harden ingesting autoscalers around task-count boundaries

[Fly-Style]

This PR:

• fix lag-based autoscaler by using taskCount from ioConfig for scale action calculations instead of activeTaskGroups;
• hardens seekable-stream autoscalers when a supervisor is configured with a handwritten taskCount outside the allowed bounds. For both cost-based and lag-based autoscalers, if the current taskCount is below taskCountMin or above taskCountMax, the scaler now returns the nearest valid boundary instead of using the out-of-range value as the scaling baseline. This keeps supervisors within configured limits and avoids inconsistent scaling decisions.

This PR has:

• been self-reviewed.
• a release note entry in the PR description.
• added comments explaining the "why" and the intent of the code wherever would not be obvious for an unfamiliar reader.
• added unit tests or modified existing tests to cover new code paths, ensuring the threshold for code coverage is met.

[amaechler]

🐻

[Fly-Style]

cc @zhangyue19921010