Attempt to make BcastUntilActxArray work for broadcastees that are containers

[majosm]

In some mirgecom examples, dt can be a DOFArray. When using _compiled_lsrk45_step from grudge (which now uses BcastUntilActxArray for h, etc.), this causes dt to broadcast down to the array context arrays of the RHS container, leading to the creation of erroneous nested DOFArrays. So it looks like we need to tweak BcastUntilActxArray to handle the case of broadcastee being an array container.

I took a stab at doing so here. I don't know if this is the right approach, but my hunch is that we should let the container's defined broadcasting behavior take over in this case (mainly because I feel like, e.g., bcast(Container()) + Container() ought to be equivalent to Container() + Container() and not do any kind of broadcasting inside). Thoughts @inducer, @alexfikl?

[inducer]

In some mirgecom examples, dt can be a DOFArray.

Could you explain? What does it mean if the timestep is a DOFArray?

[majosm]

In some mirgecom examples, dt can be a DOFArray.

Could you explain? What does it mean if the timestep is a DOFArray?

It's from Tulio's local-timestepping stuff. The dt value can vary per element.