I'd say we should definitely look into the frequency (how often?) and severity (how difficult is to fix?) of breakage if we were to change floaters/gliders, as personally I consider them analogous to fallers, except they survive falls.
Just making up some numbers here, but if it was, say, 3 levels that were broken out of all existing levels (I suspect the true number would be higher), then it wouldn't be a severe change to clean up the rules even if the levels were very difficult to fix, while if it's very simple to fix (e.g. just needs some terrain under the exit and doesn't cause major backroute issues), then a much larger number of broken levels could be tolerated. I think we should find out this information before we commit to changing/leaving the behavior.
I also think the vortex is relevant to this discussion, as it could be a viable solution for fixing some of the broken levels, depending on how the floating exit is used in broken levels.