This pull request's main purpose was to fix line 279 to correct the non-homogeneous numpy array that ensued from using joint features, in my experience. Additionally, I added typing for documentation purposes, formatted code to be more legible, and streamlined find_seq_features and _load_features_and_model by removing for loops and using more efficient functions. I tested this version of SoNNia compared to the current release, and the Q's correlated to 0.9998, so everything should be the same there.
This pull request's main purpose was to fix line 279 to correct the non-homogeneous numpy array that ensued from using joint features, in my experience. Additionally, I added typing for documentation purposes, formatted code to be more legible, and streamlined find_seq_features and _load_features_and_model by removing for loops and using more efficient functions. I tested this version of SoNNia compared to the current release, and the Q's correlated to 0.9998, so everything should be the same there.