Re-create indexes in case they are missing

If index files are missing and the data is not, `MerkleDbDataSource` should be able to reconstruct the indexes on fly.
Once the data indexes are reconstructed, they have to be re-hashed.

Start by locating MerkleDbDataSource and tracing how its data and index files are loaded. Check how missing index files can be reconstructed from existing data, then determine where the reconstructed indexes must be re-hashed. Done means missing indexes are rebuilt on the fly and their hashes are updated consistently.

Written by the indexing model from the issue text.