In each of the many-worlds it is a particle. The wave behavior is caused by the interaction of the many-worlds.
Does that explain the two slit experiment, with a single particle inferring with its own wave function? If you consider it to be a particle, then you probably need to make the interact of the many-worlds be wavelike. At that point you probably have a more complex model than admitting the photon was not a particle.
Yes
> At that point you probably have a more complex model than admitting the photon was not a particle.
Yes again. But then why a more complex model? To be able to deal with interference of complex systems of entangled particles.