[SUPPORT] Inconsistent Checkpoint Size in Flink Applications with MoR

I have an Apache Flink Application that leverages RockDB and incremental checkpoint, but it seems that every time a compaction task occurs, the checkpoint size of the application during that interval increases dramatically. Is this due to when doing compaction, the streaming job has to load the entire table? if thats the case, Is there any way it can only compact and load the current partition?

No source files, tests, or reproducible configuration are identified. Begin by collecting the Flink/RocksDB incremental-checkpoint and compaction settings and comparing checkpoint metrics before and during compaction; done requires a confirmed explanation and an actionable scope for the current-partition behavior.

Written by the indexing model from the issue text.