Split `ligo.em_bright.source_classification_pe` function

The purpose of this MR is to propose a small change which splits the ligo.em_bright.source_classification_pe function into 2: one function which reads the file and extracts the posterior samples and another function which calculates the probabilities using the posterior samples. This allows other packages (e.g. pesummary) to directly call the proposed ligo.em_bright.source_classification_pe_samples function.

This came about because I was recently made aware of a change made 8 months ago which added the tilt angles to the computation while pesummary still uses a1 and a2 (i.e. no tilt angles).