Auto-pruning database

Incremental database updates will now by default auto-prune to remove previously-indexed files that no longer exist. This can be disabled by passing prune=False to the indexing functions.

Additionally, we use a broader and hopefully more robust search to figure out what the time dimension is within a file.

More robust indexing

Index should proceed despite inaccessible files/directories being thrown by the find utility.

No payu dependence

This incorporates the restructure from #130, which structures the database in a more relational way, and no longer depends on experiments following the payu directory structure. This means that observational datasets can be easily indexed alongside regular model output. The indexing itself is also more robust, and should give a database that is more easily explored.

(Re-pushed for consistent version numbering)

Final version with old database compatibility

With the introduction of the new database format for the cookbook, a lot of old functionality is incompatible. This version is the last release of the Cookbook compatible with the old database format.

(Re-pushed for consistent version numbering)