Lots of good discussion here; I have some replies and reactions:
Sure, I will look into getting other drag-and-drop or fixed architecture contractor-style DSP units such as the Symetrix and the Biamp models. I don't think the BSS Soundweb has auto feedback filters but I would not mind someone proving me wrong.
The main application I'm thinking about is speech. Music is of course a more difficult application because various musical sounds can be mistaken for feedback. But then a speech microphone is also a singing microphone in most HOW settings, so the AFS-type device should be at least somewhat resistant to the fairly pure tones that come from whistling or flute. And it should be capable of filter lockdown during shows.
I don't need to borrow a Peavey Feedback Ferret because I have a digital implementation of one in my Mediamatrix X-Frame 88. Thanks, though!
Regarding the idea that a delayed version of the initial signal can be fed back into the original signal to reduce feedback, the problem with that is it would probably sound annoying to everyone in the audience, with various delayed versions coming out of the loudspeakers at a lower level.
Thanks for the suggestion that the DR260 gives a more satisfying experience of dbx AFS than the AFS224 unit (which has no computer connection at all.)
Thanks for the suggestion of comparing a DFR11 with a DFR22.
Regarding the width of notch filters, I have seen some ridiculously skinny ones such as 1/70th of an octave. A lot more common is 1/10th of an octave, which is a bit bigger than one piano note. Over the years I have settled on a personal preference for 1/5th octave wide notches because these are more robust in the presence of changing air temperature.
Considering the Behringer product range, I don't particularly care which feedback "destroyer" appears at the shootout. I don't think Behringer will impress anybody in any form. So you can see my prejudice.
Inserts: If a feedback fighter is inserted on a channel then I would want a clean version of that channel to send to recording and streaming mixes. Personally, I would something like this on a group mix.
-Bink