Choose minimal velocity scale to fix bc issues

[wjhorne]

This PR fixes an issue found in Nalu-Wind's VOF formulation around solid boundaries. Essentially the compression and diffusion of the interface should be zero along interfaces or strange behavior occurs. The PR addresses this issue, while avoiding issues with mesh motion, by taking the local velocity scale to always be the minimum of the local velocity and the volume flow rate.

@mbkuhn the flag approach proved to be complex due to the flag being tied to balanced buoyancy. I found this as a simpler way to tackle things and to possibly get the best of both worlds. To recreate the flag approach, we can also just check if vdot ~ 0.0 and then zero out the local velocity scale rather than always taking the minimum.

[mbkuhn]

We've already talked about this, but I'm putting it here for the record.

Though this implementation works pretty well, we've seen it be less robust than the original volume-flux-based approach. This is probably due to instances when the velocity (within the min expression) becomes preferred, and this appears to be less stable. We think the preferred approach is to incorporate the mesh velocity into a volume-flux-based approach so we can retain the volume flux base but also address mesh motion.

[mbkuhn]

we may end up needing to have something along the lines of min(volume_flux, volume_flux-mesh_velocity)