/skip does not fully clear stale non-blocking failures in prow bot comment (especially after presubmit job rename)

[kkauy]

What happened:

On some PRs, the prow bot comment lists multiple non-blocking job failures that appear stale or unrelated to the current commit.

Running /skip (and sometimes /clear) may ignore certain CI statuses, but stale non-blocking failures can remain visible in the bot comment.
In addition, /skip may not behave as expected when a presubmit job has been renamed, leaving outdated results in the comment.

This creates confusing CI signal for reviewers even when all required GitHub checks are passing.

What you expected to happen:

One of the following:

• /skip should fully clear or reset stale non-blocking failures shown in prow bot comments, or
• There should be a documented and reliable way to reset the bot comment state, including cases where presubmit job names change.

How to reproduce it (as minimally and precisely as possible):

1. Open a PR where the prow bot comment lists non-blocking failed jobs.
2. Confirm required GitHub checks are passing and merge is not blocked.
3. Issue /skip (and optionally /clear) on the PR.
4. Observe that:
   • Required checks remain correct, but
   • Stale non-blocking failures may still remain visible in the bot comment, or
   • Behavior may differ when presubmit job names were renamed.
     Please provide links to example occurrences, if any:

• Add Resource Version query and Bookmarks to thread safe store kubernetes#134827

Anything else we need to know?:

This behavior was discussed in SIG Testing Slack, where it was noted that:

• /skip may not fully clear stale bot comment results.
• The issue may be worse when presubmit job names are renamed.
• Prow comment state appears to be delta-updated rather than fully recomputed.

[kkauy]

/sig testing

[bleon-ethical]

/area prow

[bleon-ethical]

This is a known issue with how Prow updates the PR comment status (delta vs. full reset). It's especially visible when jobs are renamed or removed. A fix is already in progress in kubernetes-sigs/prow#615. Once that is merged and Prow is updated, /skip and general status reporting should clean up these stale contexts automatically.