Updated git actions workflow to include a new SPI unit test which will run through all SPIs in the full set, compute their outputs on a benchmark dataset (standard_normal.npy) and compare to a benchmark/reference calculator. In the current state, unit tests will only identify when SPIs 'break'. Differences are provided via a warning (not an outright fail) until a strategy is formulated for determining what constitutes a meaningful difference (and this is very much SPI dependent). These differences will only be reported when they exceed machine epsilon for float64.
Included a standard_normal_calculator.pkl which stores a reference Calculator object (with benchmark table) for running unit tests. This calculator serves as a standard reference and the SPI outputs were generated in a docker environment using the dependencies + python version, etc. specified in the DOCKERFILE.
Removed old docs directory and .readthedocs.yaml file
Added test_SPI.py file containing unit tests pertaining to the SPIs as described in point 1.
The following changes were made: