MAJOR: Improve test coverage by updating notebooks to use limited set of test locations/times 🗺

[robbibt]

Currently our notebooks cover many different locations and time periods. This makes it tricky to test our notebooks, as we need to keep adding new datasets to our test database here: https://github.com/GeoscienceAustralia/dea-notebooks/blob/develop/Tests/setup_test_datacube.sh

To make it easier to test our notebooks, we could:

1. Settle on a limited set of "golden" tiles that cover most of the environments we focus on in the notebooks. For example:
   • one tile in Canberra (urban, rural, forest)
   • one coastal tile (Queensland somewhere?)
   • one inland water tile (Menindee lakes)
2. Settle on a single year of data (e.g. 2020)
3. Index required datasets into test datacube for those tiles/time periods
4. Update notebooks to load data from one of those locations and time periods

This would be a large piece of work as it would involve updating the x/y/time query for each notebook. However, once this is done, we could reliably point all future notebooks to one of these test locations, and automatically have the notebook covered by our integration tests (potentially saving us a large amount of time and effort tracking down bugs and problems).

WOuldn't need to remove test datacube data - but add new golden tiles and use them going forward