[APIView] Automatically Clean Up Stale Comments in API View

There is an accumulation of stale comments in API View, especially after a library reaches General Availability (GA). These comments are often no longer relevant or actionable, as API design issues are difficult to address post-GA.

There are two aspects on this problem:
- new comments
- existing comments

## New comments
The proposal is to implement a mechanism that auto-resolves or removes non-must-fix comments when a library goes GA, while providing a way to retain or track important design discussions separately (e.g., via a special label or by opening GitHub issues for future consideration).

## Existing comments
The problem is how are we going to clean up the existing stale comments, what is a good rule to delete the existing comments? can we delete the comments that specific to revision that are related to a GA release version?

I tried to get information about how many comments are in this place, but the number is not accurate but is on the thousand, around 4,000, and there are revisions heavily impacted like [Azure.ResourceManager.ApiManagement](https://spa.apiview.dev/review/441201e06d7e48a895102a33026828c0?activeApiRevisionId=58e3c37356fe47edbaeecf2c8c68b3fd) with 253 comment, or [Azure.ResourceManager.Storage](https://spa.apiview.dev/review/552ab789297f4327bd68aed12c6db2ff?activeApiRevisionId=b5d1e2abb8bc4615bbd36af368dcaf88) with 167 comments

## Notes
There are additional considerations in here
- When API lines are changed (renamed) we need to not loss the comments that was originally tied to the elementId (token Id) - we should investigate if this is still required

The issue concerns API View comments and gives example review links, but names no source files or tests. Start by locating the API View comment model and the library GA lifecycle, then investigate how comments are tied to revisions and element IDs. Done should include agreed rules for new and existing comments, retention of important discussions, and handling renamed API lines.

Written by the indexing model from the issue text.