Even when progressive, PAL is not optimal. 3 are the cases of progressive PAL:
1) It was blended to 25: worst case, you can save less than what you could attempt to save when you have a field blended source. Admittedly, this case is luckily rare to nonexistant in my experience (I think I read about it but never found it myself).
2) A frame was added every 24: in this case, the frame could either be a duplicate or a blend of the two frames close to it. I have found this case a few times, it is the most common for progressive PAL in my experience. All you have to do is tdecimate(cycle=25) and hope tdecimate is smart enough to remove the duplicate/blended one. Alternatively you can use an ovr file in tdecimate or try a selectevery to decide which frame is the duplicate/blended one that you should be getting rid of. After you restore the 24.00fps stream like this, you can assumefps to ntcs_film.
3) There was a direct speedup to 25fps: in this case you just assumefps back to ntcs_film. You could still edit at 25fps if you wish, but in some places you might notice that the motion is funnily faster than it should be: a common case is when there's someone walking in a somewhat robotic way due to being accelerated like that. Generally however speedups aren't common in progressive
PAL because when the source did go through a speedup, it generally is a source dealt with in the 70s/80s, and at the time they generally did full field interlacing with the TV airing in mind, so they just picked the source, dropped the audio, did the speedup, and interlaced, with the dub done at the new speed. The only progressive PAL DVD that went through a speedup that I have in mind is the Italian release of The Sky Crawlers by Dall'Angelo Pictures (not that it matters, they also did the BD at 23.976, so I got that