Move hardcoded config into external config file

[rod-glover]

Resolves #86 Resolves #89 Resolves #92 Resolves #85

This PR turned out to be more worthwhile than anticipated. The move to configuration by file tightened up the code enormously. This is now a professional grade app suitable for management and deployment by people other than its developer.

TODO:

• [x] Title
• [x] Colour scale definitions in src/components/DataMap/stationColor.js
• [x] Appearance of station markers: src/components/DataMap/DataMap.js - locationMarkerOptions, dataMarkerOptions
• [x] Contents of src/utils/variables.js
• [x] Datasets and labels thereof
• [x] Layer group names and order
• [x] Backend URL(s)
• [x] Style for controls
• [x] Date selector config

This PR touches a lot of files, but the touches consist almost entirely of:

• Move config to the config file; get it via the config context (const config = useConfig())
• Code cleanup:
  • Remove all smoke tests. They deliver little to no value, and impose a maintenance overhead.
  • Remove other unnecessary files (CSS, unused components, etc.)
  • Split up files defining multiple components.
  • Add a few new utility components (e.g., VariableUnits).

[rod-glover]

Demo

@jameshiebert , I'd appreciate your feedback on the UI. There are a few minor changes to behaviour and appearance:

• Variable, dataset, and date controls are fully configurable. (Minor changes to their appearance in the present config.)
• Colour scale is fully configurable. (Minor changes to their appearance in the present config.)
• Station location marker appearance is fully configurable. (No change.)
• Station data markers always show the station location with a location marker in at the centre of the coloured marker. (This was not true before.)
• Using the layer control, you can optionally turn on markers (small dots again) for baseline and/or monthly data as a comparison. (Layer names configurable, because why not.)
• Location markers are always overlaid on coloured data markers. (Configurable.)
• Many other items, e.g., labels for variables and datasets, are configurable.