ClusterManagerTaskThrottler Improvements (opensearch-project#15508)

* [ClusterManagerTaskThrottler Improvements] : Add shallow check in ClusterManagerTaskThrottler to fail fast before computeIfPresent to avoid lock when queue is full

Signed-off-by: Sumit Bansal <sumitsb@amazon.com>

Author: sumitasr

CHANGELOG.md

@@ -49,10 +49,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 - Add prefix support to hashed prefix & infix path types on remote store ([#15557](https://github.com/opensearch-project/OpenSearch/pull/15557))
 - Optimise snapshot deletion to speed up snapshot deletion and creation ([#15568](https://github.com/opensearch-project/OpenSearch/pull/15568))
 - [Remote Publication] Added checksum validation for cluster state behind a cluster setting ([#15218](https://github.com/opensearch-project/OpenSearch/pull/15218))
-- Add canRemain method to TargetPoolAllocationDecider to move shards from local to remote pool for hot to warm tiering ([#15010](https://github.com/opensearch-project/OpenSearch/pull/15010))
 - ClusterManagerTaskThrottler Improvements ([#15508](https://github.com/opensearch-project/OpenSearch/pull/15508))
-- Reset DiscoveryNodes in all transport node actions request ([#15131](https://github.com/opensearch-project/OpenSearch/pull/15131))
-- Relax the join validation for Remote State publication ([#15471](https://github.com/opensearch-project/OpenSearch/pull/15471))
 
 ### Dependencies
 - Bump `netty` from 4.1.111.Final to 4.1.112.Final ([#15081](https://github.com/opensearch-project/OpenSearch/pull/15081))