[Remote Cluster State] Allow Remote Cluster state writes to remote use a dedicated threadpool

[linuxpi]

Describe the bug

• Today Remote cluster state write to remote happens asynchronously. This allows us to parallelize upload of multiple files and reduce overall write time to remote
• We use the PRIORITY threadpool introduced by the repository-s3 plugin.
• Remote Translog upload also uses the same threadpool. So in a non dedicated cluster manager setup, the remote cluster state writes would be competing with remote translog writes.
• This can lead to issues in case translog uploads hogs up the pool and starves remote cluster state writes.

Expected behavior
Remote cluster state write should use a dedicated threadpool to avoid any interruptions due to resource starvation