[FEATURE] Checkpoint folder data cleanup limitation

[dai-chen]

Is your feature request related to a problem?

Currently Flint pass the exact same checkpoint location path given by user to Spark. It's user responsibility to:

1. Make sure the checkpoint location is used by only 1 Flint index refreshing
2. Clean up the checkpoint folder after Flint index drop

What solution would you like?

Flint cleans up the folder when user drops index. Add this to limitation doc if no time to improve this in 0.1.1 release.

[dai-chen]

This limitation could be removed in #104.