Answering your first question, in general the selection of the temporal baseline, it depends on the stability of objects in your study area, the goal of your study, what kind of result you intent to achieve; The far period of S1, is 12 days and 6 days, so the selection short period promotes the quality of the coherence, and hence affects positively in most cases the quality of the produced interferograms.
I’d suggest using the same schedule but in shortest time interval.
Technically concerning your approach please take a look at this post,
and please take a look at this,
Selecting the reference point let’s call the technique you’d to preform is “stacking” it is still not supported nor the technique neither the selection of the reference point in the step of unwrapping in case of processing a single interferogram,
Here, I’d say that the PSI, SBAS technique using StamPS, is supported by command of selection the reference point, good background information of your study area helps to choose the right stable reference point, also the best master image, technically, please take a look at this post, and also you should be sure about the radius,
In here, if you mean the reference point doesn’t change corresponding to time your data selection, that’s great.