MintPy is another option (I have not looked in PyRate yet - there is too many options right now it is so hard to choose!).
Check my post and the MintPy wiki. More people looking at the problem = better solutions!
You can now ingest .dim files directly into Mintpy for small baseline processing.
In terms of master / slave selection - you can use the method below from:
Fattahi, H., P. Agram, and M. Simons (2016), A Network-Based Enhanced Spectral Diversity Approach for TOPS Time-Series Analysis, IEEE Transactions on Geoscience and Remote Sensing , 55 (2), 777-786, doi:10.1109/TGRS.2016.2614925.
This way you are creating a redundant network - in my testing in a rural area it looks like up to n=4 makes sense with Sentinel-1.