DAGMC: access geometry ID spaces from the Python API

[pshriwise]

I was staring at the DAGMC H5M format today and I think that determining the ID space in the file shouldn't be too difficult to do, contrary to my original opinion. Being able to use these ID spaces in the Python API might allow us to remove the auto_geom_ids option from the DAGMCUniverse and gain some transparency on what IDs are available when creating CSG cells to avoid conflicts before runtime.

[shimwell]

Also keen to see this feature in openmc. I think we coded up something together in this package a while back that got volumes, materials or volumes and materials. That last combination of volumes with their associated materials would be super useful to see in OpenMC with h5py. Keen to see this feature

[ebknudsen]

Please correct me if I write something wrong here: The h5m-format has the notions of vertices,curves,volumes, and groups, right? (and more but we don't really use that). Groups are volumes with the same material tag, whereas volumes correspond to OpenMC cells. If the material tag could be allowed to mean something different than material we could address it.

[pshriwise]

Can you elaborate on what you mean by "something different"? What would it's meaning be instead?

[ebknudsen]

Sorry, I wasn't clear. I simply meant that the tag would instead mean a universe. Or be interpreted as such. This could be done load-time perhaps. Through a dictionary or similar mapping materials to universes.

[pshriwise]

Ah, yes yes. That's what I was thinking too. Though this line of thought might be more related to the google doc linked in #2549