There are two main observations. One is If we expand the OLG, the phase turn begins later. The other is it turns more viciously later on. With the 1u capacitor overall phase is smoother, does not skip zero point phase magnitude before it. Amplifiers reflect heavily on their hi-band behavior of stability and extension, plus what happens lower has its impact too. Not to mention after real parts, wiring, and layout, watching those sims is mainly academic. What counts for an experimenter is to keep stable and be happy subjectively. I can not possibly glean such refinements from sim graphs beyond alarming. So I prefer to recommend the conservative values for start. We don't know what happens in different layouts and load combinations yet. Looks like it can afford smaller C1 for now, possibly opening up also, and that's good practical feedback. Keep on reporting.