My first take is similar to yours, that it might be an error. However note that for this section the commentary explicitly acknowledges that the zones are larger (see the last 2/3 of the first paragraph C29.4.3). Also I note that the figure in the commentary C29.4-1 repeats the dimension 2h (not 0.2h). Therefore although it may seem coincidentally unusual (i.e., by a factor of 10 - did someone drop a decimal?) my interpretation is that the intent is as-shown (i.e., 2h, not 0.2h). Also I'd suggest that no, Figures 30.x are not more applicable that 28.x because in 29 you're looking for MWFRS, not C&C.
For the second question, I don't see a problem in the dimension being a function of h; this actually seems appropriate to me. However I'm probably missing your intent/usage of 'solely', are you suggesting another parameter (e.g., B or L) or a limit? Either way, yes I believe zone 3 could cover the entire roof if/as-determined by testing that has been done and yes there is testing that has been done (see Commentary). In short, without wading through the relevant literature, I'd take what's presented at face-value. Sorry though, only an ASCE person can truly tell you if this is an error.
Hope this helps,