Add hash caching

[eywalker]

Currently a file hashing is computed every single time a file is encountered, even if the file's hash has been previously calculated. While this may be good for being sure about the file content reproducibility, it is quite prohibitive with respect deeply nested pathset hash calculation.

Implement a caching mechanism in a modular manner (a caching object that can be passed into the underlying data store). A good candidate for the first caching backend would be SQLite.

[eywalker]

Addressed in #18