Support for:
* Listing datasets in the global data project stack with `ls`,`list`
* Showing the content of datasets with `show`
* Manipulate the global data project stack with `stack list`, `stack push`, `stack pop`
* Help with `help`/`?`

Includes REPL completions, help text and some tests.

Doesn't yet include support for adding new datasets to existing projects.

Now only need to update one place when the list of test datasets
changes.

While it's helpful for direct users, this can show up in unexpected
places when DataSets is loaded by another module. So probably best to
disable this.

Various cleanups

This allows modules which provide data storage drivers to be
automatically loaded when DataSets itself is loaded, providing a
declarative data environment workflow while bypassing world age
issues we'd get from loading these on demand.

It seems somewhat unclear whether interfering with code loading
like this is a good idea, but I think only time and some experience
with this mechanism will tell...

This is the obvious hierarchical separator, and consistent with having
URNs as an expanded form for dataset names (see RFC8141
https://datatracker.ietf.org/doc/html/rfc8141)

The only real downside is that it uses some syntax which might be nice
for a path component, but we could potentially use something like
namespace/name:a/b/c.