The thing is, if you’ve been able to revert to the previous version, everything should be back to the way it was — upgrading to CM4/Grbl 1.1 and then reverting to CM3 and Grbl 0.9 should get one back to exactly the same state.
I’ve got your file in the support queue and will be trying to run it presently. If it runs successfully w/ the spindle off, then it seems to me that EMI is the most likely culprit. I wish we had money to set up every person doing support w/ every machine variation, including third-party spindles, but that’s just not in the budget — if you’re going to reach beyond the standard supported configuration you’ve got to bridge that yourself and be self-supporting on the non-standard aspects of the machine.
One possibility here is that you didn’t revert things to exactly the same state — please check your table extents and steps / mm — is it possible that it is alarming on trying to move out of bounds? (I had to reduce both files to get them to run on my XL)