Great question GSD! A definitive answer to your question is some years away. I'll make a few observations and what I think is likely to emerge as more data becomes available.
1. One thing to remember is that PACE-B was concluded over 5 years ago in a different county and while the protocol looks similar it may not be identical to current treatment/hardware being used here today.
2. That said, there does appear to be a small, but real increase in late GU complications with SBRT as you've noted.
3. As Ben pointed out, all SBRT isn't equal. "Regular" SBRT typically uses 5 mm margins except posteriorly where 3 mm is common. Margins in PACE-B were 4-5 mm except posteriorly where they were 3-5 mm. MR-Guided SBRT is typically uses 2 or 3 mm margins and is the most likely mechanism for any benefits over regular SBRT (i.e., not the adaptive planning or continuous monitoring).

My best guess, which I posed to the GU section chief at arguably a top U.S. cancer hospital, is that when all the data is in, we will know that SBRT (Cyberknife, Proton, etc), while more convenient, does carry a small increase in late GU complications compared to moderately hypo-fractionated treatment of 28 sessions. With MRI-guided RT (Electa or MRIdian) the complication rate is likely to be equivalent to 28 sessions while maintaining the convenience of only 5 sessions. FWIW, the "top name" radiation oncologist agreed and said that's also what he thought.

Therefore, you won't be surprised that when I'm asked for a recommendation about SBRT, my standard response is if there is a choice between 28 sessions and 5 sessions on a standard machine, AND you have the time, I'd take the 28 sessions. If you have access to a MRI-guided machine, then I'd go with the MR guided strictly for the convenience.

FYI, I had 2 mm margins on a MRIdian machine for favorable intermediate disease.