To provide a means to be able to share the underlying database, while
remaining (somewhat) anonymous, we can introduce a `db anonymise`
command to remove any identifiable information.

We produce an implementation of a SHA3 hash ourselves as mentioned in
[0] until upstream support exists. This includes some tests to validate
that the output from our function matches the SQLite implementation.

We can add additional scrubbing for fields that may contain organisation
names, but not do it by default as it can remove most of the benefits of
having the data at all.

This requires a little more widespread a change to get these working,
and may indicate the way our database schema is setup up may be starting
to need rethinking.

Closes #42.

[0]: https://gitlab.com/cznic/sqlite/-/issues/139