IISPH is not correctly considering velocity of moving boundaries

[efaulhaber]

With #751, the predicted velocity of boundary particles is defined as zero:

@propagate_inbounds function predicted_velocity(system::BoundarySystem, particle)
    return zero(SVector{ndims(system), eltype(system)})
end

For moving boundaries, this should instead be the prescribed velocity at the next time step.
There might be other instances where the equations have to be changed to correctly support moving boundaries.