recreate compacted boltdb files from compactor when they are more than 12 hours old

[sandeepsukhani]

What this PR does / why we need it:
This PR changes the compactor code to re-create compacted boltdb files using bbolt.Compact for 2 reasons:

1. When files are deleted, boltdb leaves free pages in the file. The only way to drop those free pages is to re-create the file.
   See database file size not updating? boltdb/bolt#308 for more details.
2. boltdb by default fills only about 50% of the page in the file. See https://github.com/etcd-io/bbolt/blob/master/bucket.go#L26.
   This setting is optimal for unordered writes.
   bbolt.Compact fills the whole page by setting FillPercent to 1 which works well here since while copying the data, it receives
   the index entries in order. The storage space goes down from anywhere between 25% to 50% as per my tests.

The compacted dbs would be re-created when:

1. We have just a single file in table.
2. File was compacted more than 12 hours ago to avoid re-creating the files too often while the file is still being modified by compaction or retention.
3. We have not already re-created the db since re-creating again would not have any benefits.

To simplify the code, I have also changed the compactor code to drop the requirement of having more than 4 files for doing the compaction.

Checklist

• Tests updated
• Add an entry in the CHANGELOG.md about the changes.

[cyriltovena]

LGTM

This file might need some refactoring it start to be a bit too complex for what it does.

[sandeepsukhani]

LGTM

This file might need some refactoring it start to be a bit too complex for what it does.

Thanks for the review! I am taking care of it as part of per user index changes.