Interesting. Looks like there's frequency compensation present on the CC and CV opamps. The design is a little weird though as the entire thing is referenced to the adjustable rail after current sampling rather than a low impedance ground. I'd start by trying to slow the driver transistor down T1/T101 with a bit of miller capacitance and see it that responds.
Edit: Previous attempts to design power supplies myself have always resulted in a monumental effort on stability! Diagnostics usually involve fecking around for hours measuring sampling vs control voltage phase shifts and trying to get that loop stable at all frequencies. LTspice only gets you 80% of the way there. One I did which was a simple diff amp with a couple of 2n3904's was only stable when it was a warm day. Fixed with a bit of heatshrink over the two transistors!
Makes you appreciate the effort HP put into their linear supplies over the years even if they did feck up a bit with the E36xx
Edit 2: actually a mini rant. You see loads of crappy power supplies described all over the internet and in electronics magazines. They worked for the author, precisely once under one set of load conditions.