feat: data initialisation, criteria and assessment parameters refactor

[PeterBaker0]

Missing criteria

• adds wave height and period
• makes all criteria params optional
• uses the regional assessment criteria ranges explicitly to build out the query param dict
• merges user provided ranges with the default from the dataset and warns if ranges are OOB

This has the effect of

a) allowing users to run jobs only restricting criteria of interest
b) ensuring regional and suitability assessment caches are shared where they should be, but invalidated if defaults change (by ensuring the query param dict passed to the hash function includes ALL criteria, not just those specified in the request)

Criteria, data initialisation and assessment params refactor

Rewrites the data initialisation, caching, criteria bounds etc all into a unified data structure which is fully typed. Did use structs all the way through but found that naming criteria in the structs ended up being over-engineered compared to having const global arrays/dicts of available criteria.

This has lots of side benefits like fixing regional assessment file caching, properly merging default ranges and user ranges (including into hash), and making criteria easy to add/exclude.

The process to add new criteria now should be easy i.e.

• define the criteria metadata
• specify which regions can use it

There is also a lot of additional runtime validation on data loading relating to presence of necessary fields.

[ConnectedSystems]

Very minor comments, feel free to merge once addressed