Infofiles may be out of sync when using dst_tighten_bounds in chained runs

[dodamih]

When dst_tighten_bounds is used in chained subchunkable runs anywhere that's not the first run, the workers that have already accessed the dst VolumetricLayer will have the non-tightened bbox in their cache.

We've never had a way to push change notifications to workers, but have gotten away with it so far because the assumption was that the infofile would never change in the middle of a run. Now that tightening bboxes is a thing, we should figure out a way to make sure that updated infofiles get propagated.