Oh Dave, IMHO perhaps you should have been very explicit that you were talking about AC *trigger* coupling, not AC *signal* coupling. I can't tell what proportion of people, especially in the YouTube comments, actually understand the distinction (I sure didn't know AC trigger coupling even existed until watching this video; I'm struggling to contrive applications for it), and I see at least a few people who quite clearly don't and think that AC coupling is broken. Then again, if you were to be completely clear about every little thing, your videos would be 24 hours long, so I do understand that this is a tricky balance.
Also, more importantly:
The issue with the clock signal not generating the eye diagram you expected is not a bug -- the holdoff is a user-defined value; if that value is such that the next edge after a rising edge + holdoff delay is another rising edge, then you'll lock onto those rising edges and get rising edges all day long. A randomly chosen holdoff value has a 50/50 chance of either A) finding alternating rising/falling edges (good eye diagram), or B) finding only one or the other forever (apparently "ignoring" your request to detect both edges). By modifying the holdoff value on my DS2202, I can easily transition from A, to B (holdoff half a period longer), to A (holdoff a full period longer), to B, etc etc -- exactly as you would expect. I'd expect even analog scopes (do they have a holdoff knob?) to behave the same way; if they don't, that needs explaining. Generating eye diagrams on actually interesting data signals works fine on my DS2202 more-or-less regardless of holdoff value since the random data randomizes which edge is found next. I haven't explained myself to well, but hopefully y'all can see what I mean -- in short, that particular "issue" is totally not a bug at all, it's doing precisely what you asked for.
EDIT: To be clear, I'm not exactly defending Rigol, my opinion is:
-- Oscilloscope "ignores" request to trigger on both rising and falling edge: Not a bug at all
-- AC trigger coupling issue: Really weird bug that needs to be fixed, but the impact is miniscule compared to what it would be if AC signal coupling was broken.
-- The 5us thing: Bad bug that needs to be fixed.