feat: segment availability improvements

[jtuglu1]

Fixes #18764.

Description

During move of a 1x-replicated segment from historical A to historical B, the coordinator can currently race and drop a segment temporarily from the timeline.

This occurs through 2 primary ways:

1. The SegmentLoadQueueManager.moveSegment operation calls unmarkSegmentToDrop and then dropSegment which can race with readers.
2. The coordinator currently receives information about what segments are loaded where through an inventory view (ServerInventoryView) and historical node callbacks (HttpLoadQueuePeon). Since these historical callbacks typically always arrive first, the code sums up load data from both sources and then does a diff. While this works if the HttpLoadQueuePeon is present in the queue (to cancel out the new pending load on historical B), before this fix, the peon deletes the holder the moment the server acks. So the DROP left the queue while the inventory still showed the segment loaded.

Changes

1. Ensure we do the mark to drop transition atomically (SegmentMoveConcurrencyTest::testMarkToDropGraduatesToQueuedDrop).
2. Make sure we hang on to callbacks until inventory catches up.

Release note

---

This PR has:

• been self-reviewed.
  • using the concurrency checklist (Remove this item if the PR doesn't have any relation to concurrency.)
• added documentation for new or modified features or behaviors.
• a release note entry in the PR description.
• added Javadocs for most classes and all non-trivial methods. Linked related entities via Javadoc links.
• added or updated version, license, or notice information in licenses.yaml
• 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.
• added integration tests.
• been tested in a test Druid cluster.