Datacube-explorer

Minor fixes.

- Support for geoalchemy-2's recent 0.9.0 release (288)
- Fix for S3 URL construction (287)

Downloadable datasets

The dataset page now shows expandable links for the individual files in a dataset:

Note that `s3://` urls are converted to `https://` and you may want to configure the default bucket using `CUBEDASH_DATA_S3_REGION` in the [config](https://github.com/opendatacube/datacube-explorer#how-can-i-configure-the-deployment)

(via alexgleith in 280)


Other changes

- Added CORS headers by default to api endpoints (via andrewdhicks in 276)
- Minor improvements to repository dependencies, linting.
- New Docker build process

Minor improvements.

- Add an option to `cubedash-gen` to set a minimum amount of time to be scanned:

--minimum-scan-window TIME-LENGTH
Always rescan this period of time for
dataset changes, even if the refresh tool
has run more recently.

This is useful if you have a system where
rows are not made visible immediately to Explorer,
such as a sync service from another location.

Example values: '24h' or '1h30m'

- To match expected behaviour, the `--force-recreate` flag now forces product extent recreation too, even when no changes are detected. This more closely matches people's expectation that it "forces everything to be recalculated".

Now Incrementally Updating

This release includes a rewrite of the summary generation (`cubedash-gen`) to be _significantly_ faster at updating, and to behave in a more predictable manner.

It now applies an incremental diff of changes from the upstream ODC, rather than requiring a full rebuild.

(260, 269, 266)

Other Changes

- The DE Africa theme has been updated with the new DE Africa branding (267)
- Improvements to the `cubedash-page-test` command, to report slow pages and errors (272)
(see `cubedash-page-test --help`)
- Improved load speed of the storage page on environments with higher DB latency. (268)
- Minor Stac compatibility fixes (265)
- Various small improvements to the repository, developer setup and sqlalchemy 1.4 compatibility.

Migration

This release includes **schema updates**. You need to update your index by running the `cubedash-gen --init` command before upgrading your Explorer web instance. The updated schema will still work on older Explorer web instances.

You will see `schema.applying_update.*` messages in the log when applying the updates. They are small and can be applied while clients are still active.

**New ODC Requirements**

Last year Open Data Cube [added](https://github.com/opendatacube/datacube-core/pull/951) optional columns to record when datasets were changed. These columns are now mandatory for Explorer 2.5, as it uses them to scan for changes.

`cubedash-gen`'s schema update will try to add these itself, and tell you how to do so manually if it does not have sufficient permissions.

**Less Force**

The `--force-refresh` option on `cubedash-gen` is now strongly discouraged, unless you are manually deleting rows from your ODC database.

Many users were using this previously to update their Explorer, but it forces the entire collection to be rebuilt and so will still be very slow.

Minor style fixes:

- Improve link contrast on dark panels (they could be difficult to read on some screens).
- Restructure the product and metadata-type list pages for more visual consistency and readability. They looked a bit messy before.

- Expand the about-page API docs
- Include a `csv` button on the arrivals page.