It's classical ray optics that fails in the path-not-longer-than-wavelength regime. Classical wave optics works in that regime. Where classical techniques fail is at low brightness (because you start resolving individual photons).
Yes, but in classical wave optics you're no longer talking about "paths" as they appear in path integrals. Classical wave optics is basically quantum wave optics without the discreteness of detections, i.e., interpreting the wave as a straightforward EM field intensity instead of as a probability amplitude for detecting a photon.
