Interesting that a bad JFET can effect the drift and noise in the ohms mode. The crude function was still working, so the JFET was not totally dead, more like either excessive gate leakage or the threshold dirfte too much to be boarderling (which I would consider not very likely).
One may want to check the protection part ( the high voltage PNPs) too.
The J112 may not be such a bad choice for the ciruit, especially if the controling OP still has enough headroom for control.