I don't think it's a problem.. the flux likely gums up underneath, and is a pain to clean. If it's washable, maybe avoid vias underneath so there's less exposed metal to corrode. Errant solder balls stuck to the PCB or package, don't really matter, and most will coalesce into the pad. Somewhat less paste might be used, given the excess pad length, but it also still needs to span most of the length of the pad else there won't be enough flux and spreading ability to tin the whole thing for both components (i.e. don't make a pile at just the heel or toe). So, probably a bit of negative paste expansion would be the most you'd bother with.
Note that heel can be set to minimum or even somewhat negative, for the narrow body part -- most assemblers are only interested in one fillet, which would be toe in this case, and the heel then is optional. Note a severely negative heel will emit a "pin on soldermask" error from most DFM tools, which you may or may not wish to honor.
I think pin-on-mask probably doesn't matter too much here, as the radius of the solder blob (meniscus) on the pad should be pretty generous for SOIC, thus even a fairly tilted component should still dip into the blob and make a connection. (Or if the pin melts first, then that blob onto the pad, whatever.) I'd be more worried about that for fine pitch (TSSOP etc.?) where the pad width is much less, I think?
I don't think I'd go so far as to halve the joint face area (effectively dimension L, but in the solder joint, from pin tip to pad heel), but you can still see where it takes you with minimal heel or a little under, with respect to body sizes, and tolerances.
Tim