Much of the early work regarding increased rake angle (with trail following as consequence) was simply an attempt to cure wandering and head-shaking at speed due to lack of stiffness at the steering head. It was a crutch, and when they finally figured out how much 3-dimensional area was needed to insure sufficient rigidity, the rake requirement went down.
Testing an early (and many current) drag bike by inserting a bar through the steering head, and another through the swing-arm pivot would show that the bars can be rotated by far less effort than your nerves can stand.
Foale (paraphrasing) views trail as more important, but the large numbers frequently seen are not necessary if, again, the frame works. Not an accident that 200 mph road racers do quite well with 24° & 3".
Regarding "flop": in order to reduce flop to zero (viz., where the front wheel will not "fall over" to either side under vertical load) the rake angle must be zero. With any caster angle at all, the axle at 0° steering angle is elevated above its vertical position at any other angle. The axle climbs when the wheel is centered.
The relevant question is, IMHO, "how much angle produces sufficient axle drop (at a steering angle that will be in use) that stability is compromised?". I don't know the answer, and this is more complex that it may appear since large diameter wheels are less affected.
More comments:
http://victorylibrary.com/brit/chassis.htmI have a related question: omitting the effect on frontal area and streamlining in general, I'm puzzled about the effect of CG height in a bike.
A low CG is normally considered a good thing for most purposes, but in terms of degrees of camber needed to transfer weight for steering correction isn't a high CG an advantage here? A bike with a CG at ground level would need to incline to horizontal to make any corrections.
It seems to me that the less disturbance to the camber angle the more stable the bike will be?