It wasn't exactly the DigiFarm signal.  It was all VRS/VBN delivery signals.  The problem happened with MNCors, IADOT CORS, WISCors, Trimble VRS. It didn't happen with every 372 in every situation which made it hard to track down.  No one still knows "why" other than there was a firmware released to supposedly fix this.  DigiFarm got the brunt of it because, quite frankly, they have more users running.

If you look at the release notes for firmware 8.3 on the 372 released September 2018 one of the bug fixes was

"Problems Addressed:
  - Improvements to VRS stability. Fixing a condition in which the position
    would drop to DGPS, causing the Autopilot system to disengage.
  - Fixed a bug in position rate configuration via AgRemote.
  - Various other minor bug fixes and enhancements. "

It is my thought that the Nav would get "confused" and it would output bad data to PP.  What I was told at 10hz it would step forward 9 steps and where the "blips" were the Nav would output a "step back" and then start stepping forward again 9 steps.  So every once and a while it would think it was going in "reverse" or had stopped.  Hence, the goofy mapping.

Taking the position directly out of the 372 bypasses the Nav.

Bad data in, bad data out.

Disclaimer:  I don't own or run a 372 with PP but my Father in Law does so that's my experience.  I don't know if v8.3 actually "fixes" the problem or not.  It's just in the release notes that the stability with VRS systems is improved.  I do think you have to update the Pro700 at the same time to utilize the new Xfill frequency coming in June.  At least my FIL did.