perf: Speed up VersionedIntervalTimeline operations

[jtuglu1]

Description

We have observed severe performance degradation in the VersionedIntervalTimeline when ingesting datasource at the MINUTE segment grain. Due to query runners needing to acquire an exclusive lock here on the query hot-path, if frequent segment add/remove callbacks that touch many intervals occur, the volume + latency of each of these timeline operations can cause the BrokerServerView lock to stall indefinitely, preventing queries from running. This giant lock in BrokerServerView is IMO another issue that should be dealt with in a separate change (at the very least, should change this lock granularity to per-datasource).

This patch speeds up a few timeline operations to significantly increase timeline throughput to minimize contention on the BrokerServerView lock.

Results

Params: numInitialRootGenSegmentsPerInterval=10, numNonRootGenerations=1, useSegmentLock=false

benchLookup

Granularity	Intervals	Baseline (ops/s)	Optimized (ops/s)	Speedup
DAY	365	401,045	1,704,230	4.2x
HOUR	8,760	19,265	960,835	49.9x
MINUTE	525,600	68	243,950	3,588x

benchRemoveWithReplacement

Granularity	Intervals	Baseline (ops/s)	Optimized (ops/s)	Speedup
DAY	365	581,271	662,742	~1x (no visible change)
HOUR	8,760	19,427	19,136	~1x (no visible change)
MINUTE	525,600	45	264	5.9x

Release note

Speed up VersionedIntervalTimeline operations

---

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.

[jtuglu1]

I think #19138 could solve this problem more cleanly.