Refector how to process feature names in order to make it interpretable for downstream analysis

[axiomcura]

These dictionaries are so deeply nested - consider adding more in depth documentation outside of each utils file to explain the structure

Originally posted by @MikeLippincott in #27 (comment)

[axiomcura]

After many attempts, this now seems highly unlikely. CP features are extremely abstract, and it is very difficult to extract meaningful biological interpretation from them. I thinkwe should shift the focus toward identifying which biological compartments are affected. We can then track changes within the on-morphological features per cluster by calculating EMD scores at the channel level and potentially at the compartment level as well.

This is an idea, I'll take a look at this later.

[axiomcura]

Looking at this again, we decided not to proceed with this step. It remains challenging to discern the underlying biological processes within the morphological space, as many features are redundant and noisy. Moreover, interpretability is still limited due to the lack of a clear expectation of the biological signals represented in this space.