Implement compaction #4
Description
MangoDB data files are inefficient stores of data. Every update to an existing row is implemented as a new line at the end of an active file. This means over time we're wasting valuable disk space with stale data.
Implement the compaction algorithm which runs with a configurable frequency that would read all the data files, create a merged data file with only the latest values for a key, and then deletes the stale data files.