Switch from v0.3 to latest git version of C++ code, requires a workaround to fix the bps column of segmentation output.

Decrease cutoff for adding an extra SCE from 20Mb to 5Mb. Should help avoiding wrong inversion calls towards the end of a chromosome.

Run StrandPhaseR separately for each segmentation parameter set; this is necessary since the segmentation calls affect strand states / SCE calls.

Pipeline update to use the new SCE detection and segmentation selected (implemented in utils/detect_strand_states.py)

Add means to narrow down the number of considered genotypes (a bit less drastic than forcing biallelic); apply haplotag probabilities only after that