Compaction task reattempt fails with FileAlreadyExistsException

Issue occurs when compaction attempt is relaunched after first task attempt failure due to preemption by Scheduler or any other reason.

Since _tmp directory was created by first attempt and was left uncleaned after task attempt failure. Second attempt of the the task fails with "FileAlreadyExistsException" exception.

Caused by: org.apache.hadoop.ipc.RemoteException(org.apache.hadoop.fs.FileAlreadyExistsException): /warehouse/tablespace/managed/hive/default.db/compaction_test/_tmp_3670bbef-ba7a-4c10-918d-9a2ee17cbd22/base_0000186/bucket_00005 for client 10.xx.xx.xxx already exists