Add SolarSTEPS forecast functionality

[KristianHMoller]

This PR adds the functionality to run a SolarSTEPS forecast as an alternative to the Probabilistic Advection Forecast already implemented.

The external dependency was already included: https://github.com/EnergyWeatherAI/SolarSTEPS
Reference paper for SolarSTEPS: https://www.sciencedirect.com/science/article/pii/S030626192301139X

Dependency: The implementation builds upon the ensemble functionality, so this must be merged first: #17

[KristianHMoller]

Implementation question 1: Do we need the functionality to run both forecast methods at the same time?

[KristianHMoller]

Scientific question 1: Which values to choose for the input til SolarSTEPS:

TARGET_SMALLEST_RESOLUTION
ar_order=1,
probmatching=True,
norm=True,
local=False,
noise_kwargs={
    "noise_win_size": 90,
    "noise_std_win_size": 15,
    "noise_method": "local-SSFT",
},
norm_kwargs={"extra_normalization": True},
verbose=False,

[KristianHMoller]

Scientific Question 2: Is it meaningful to run SolarSTEPS in deterministic mode without noise?
EDIT: Answer: Currently, SolarSTEPS does not actually support noise_method = None, so it cannot run without noise. I.e., this only works in ensmble mode.

[KristianHMoller]

Scientific question 3: How to determine optimal number of cascades? Depends on resolution of the input presumably?