Prototype for a new caching mechanism, where all cached files would be under a
~/.beancount/ directory. This would encompass:

- Caching of the individual files for the parser.

- Caching of the price source lookup tool (bean-price)

- Caching for the shell's commmand-line history.

- Caching for the conversion of downloaded files during ingest.

The goal is to provide a consistent set of caching option which allow the user
to ignore, clear or specify the cache with similar option names. Right now the
options are per-tool and with inconsistent names.
