the perpendicular baseline is estimated for SLC products and can change after application of the orbit file. This is what we observe in the Stack Overview 170.92 (before) and 172.18 (after).
The difference between SNAP (172) and snaphu (215) is interesting and I do not have an explanation, so far. But I would be interested if this has an impact on the unwrapping result. So if you want to copy the snaphu folder, change the BASELINE value to 172 and run it again, we could compare if the results.
Maybe someone knows why the value used by snaphu is different.