Solar coords

[felix-e-h-p]

Decoupled solar coords - potential inclusion for further tests here, yet not sure if entirely necessary - updated tests relative to changes none the less

Relates to #179

[Sukh-P]

Thanks for making a start on this, think the logic of the code needs to be double checked and still more decoupling of the solar coordinates from GSP/site

[Sukh-P]

@felix-e-h-p I think this looks good and pretty much ready to go but I think that perhaps it still might break things on the site side, namely I am not sure this function will work now, so would it be possible to add a test which generates a site sample with the new solar configuration and then pass it through this function to see whether it gives you a numpy sample with the correct keys (namely a key that relates to solar), I think it won't and that function will need some tweaking, thanks!

Also there definitely will need to be some changes in PVNet alongside this as you can see here https://github.com/openclimatefix/PVNet/blob/main/pvnet/models/multimodal/multimodal.py#L347-L351 it relies on a naming convention that we want to move away from

[felix-e-h-p]

Cheers for the feedback.

pvnet_uk.py now has a solar config object with time parameters instead of borrowing from GSP settings (also updated relevant variable namings RE prefix).

site.py likewise has a solar config object with related timing parameters - hopefully properly handling site timestamp dimensions. t0 parameter is now passed from _get_sample method to process_and_combine_site_sample_dict so that the initial timestamp can be used for calculating the solar position timestamps accurately (no mismatch here).

Site tests updated accordingly - two additional tests.

[dfulu]

@felix-e-h-p thanks for the changes.

There is some duplication in these lines which should be removed. The code doesn't use datetimes, lon, or lat created in the if target_key == "gsp" block

Also whilst we're here, it might be best to strip out the target_key param from that process_and_combine_datasets() function. That param is no longer used

[dfulu]

This couples the solar coords to the target datetimes. It also means the solar config doesn't reflect the datetimes of the solar coords. This should be removed

[felix-e-h-p]

Wasn't sure with this, cheers. Now have created a dimension for solar position data - ensures solar coords are determined solely by config params.