Fix numSegmentsKilled metric

[abhishekrb19]

Prior to this patch, the KillTaskReport contained the numSegmentsKilled metric, representing the size of unused segments nuked from the metadata store, rather than the actual number of segments killed from deep storage. For instance, there may be cases where an unused segment won't be killed from deep storage if its load spec is in use by another segment. In such cases, only the metadata will be cleaned up. This patch contains a test for this scenario.

Alternatively and at the very least to avoid any confusion, we can capture both the number of nuked segments from the metadata store and the number of segments killed from deep storage as separate metrics in the KillTaskReport.

---

This PR has:

• been self-reviewed.
• added unit tests or modified existing tests to cover new code paths, ensuring the threshold for code coverage is met.
• been tested in a test Druid cluster.

[zachjsh]

LGTM, question: in this case, how does the orphaned segment whose metadata was deleted but data remaings in deep storage get cleaned up from deep storage, does it ever? Or is this a bug?

[abhishekrb19]

LGTM, question: in this case, how does the orphaned segment whose metadata was deleted but data remaings in deep storage get cleaned up from deep storage, does it ever? Or is this a bug?

AFAIK, the orphan segments need to be manually cleaned up from the deep storage.