you mentioned never doing measurements facing PJ. I think that couldn't be said for all home cinema setups.
I know some calibration experts prefer doing direct readings, in case of a low gain screen and less fL (because of lamp aging, or generally not so bright PJ's), or having a PJ with very dark black levels.
Especially the last point is very important to think about faced2PJ or faced2screen, because of the probes raising inaccuracy at low measurements. So I think, if using very powerful SW tools like LS we also have to feed them with most as possible accurate data at all levels.
It's true, measurements with the probe faced to screen is best, no doubts!
At least it's my experience (JVC), that measurements faced2PJ - with consideration of the screen colour shifting behaviour - is no problem. I always take care about the maximum light specification, the probe can work with.
One other aspect: The UHP lamp is very instable, even >1h runtime after power on. So performing fast measurements is also a benefit to get the whole profiling done in the shortest time as possible. And as more light a probe can see, as faster and most accurate it can work.
At the moment I calibrate my PJ with f2PJ, using the given screen offset (x=0.0019 / y=0.0030) of my white gain 1.0 screen. The results are very good.
Of course, with that method I do not involve the room's reflection light, what have to be also a point of thinking about. If a home cinema room is not light controlled, then off screen measurements should be done.
Just my 2ct.