I just did a DPR search and found 547 cases of " pe " in the MN and 1235 in the AN. And many elisions are not even marked with pe, just with ...
It's interesting to think about, though. I recently made a suggestion herehttp://pali.sirimangalo.org/forum/50/feature-requests-for-digital-pali-reader
that the DPR include frequency information along with word analysis. Forgive me if it's a bit off topic, but are you interested in using the data for prioritizing learning?
The DPR allows having different sets of texts installed, so perhaps if you moved forward with the idea, you could create an alternate set. I would recommend somehow noting what was reconstructed.
I think that the digitized BJT edition sometimes elides things slightly differently than the CST, if that's any help.