propagating hdf5 version update to CI

[bam241]

This follows #976.
now updating the version of HDF5 used in the CI.

[bam241]

I don't understand how this was working before

[bam241]

btw it appears that we build more Docker container than we use in the build_test.yml (build matrix are not the same...)

[bam241]

I think it would be nice to have the matrix existing in an other file, and pull it, but I didn't find a easy/clean way to do it :)

[gonuke]

I think it would be nice to have the matrix existing in an other file, and pull it, but I didn't find a easy/clean way to do it :)

I agree that this would be great!

[gonuke]

I want to make sure we use the right container for testing so that we don't get false positive results.

[gonuke]

I'm a little concerned about this, because it means building DAGMC in a container that already has DAGMC installed.... The moab descriptor should be from a version of the build that stopped at MOAB

[bam241]

there is no /moab container:

[gonuke]

there is no /moab container:

hmmm... Maybe that's another inconsistency that has creeped in....

[gonuke]

Let's remove the moab again and review the tests carefully to see that they are using the right build for testing.

I think we should maybe go through and delete all the old containers from the GHCR, carefully document the CI process, and start fresh.

[bam241]

from one I see in the Dockerfile we need to target external_deps

[bam241]

@gonuke external_deps descriptor seems to be ok, and it is the stage before DAGMC installation in the Dockerfile.

[gonuke]

Thanks for the follow up here @bam241