Fix functor system with ExternalProblem-derived classes

[nuclearkevin]

This PR also modifies a test to ensure we hit this in the future.

Closes #1269

[nuclearkevin]

Fighting a DAGMC build failure (CMake can't seem to find HDF5 in CIVET). We'll see if this works or not...

[nuclearkevin]

MultiAppNearestNodeTransfer is depreciated in MOOSE and throws an error if used. MultiAppGeneralFieldNearestLocationTransfer is the new version of that transfer, however it results in a diff which necessitated a regold.

Still don't know why DAGMC is failing to find HDF5 from MOAB. There might be an issue with the CIVET recipe; thoughts @aprilnovak?

[aprilnovak]

MultiAppNearestNodeTransfer is depreciated in MOOSE and throws an error if used. MultiAppGeneralFieldNearestLocationTransfer is the new version of that transfer, however it results in a diff which necessitated a regold.

Can you check that the new output file looks physically correct? (no weird nodal values?) you may need to increase the number of time steps to let it run until reaching pseudo steady-state.

As for the CIVET failure, I have no idea - nothing has changed with the test recipe and we just merged Walid's PR yesterday (which ran the test suite fine). Can we narrow it down to the MOOSE update? (e.g. could try separating out the transfer update from this PR into its own)

[nuclearkevin]

Can you check that the new output file looks physically correct? (no weird nodal values?) you may need to increase the number of time steps to let it run until reaching pseudo steady-state.

Left is the new transfer, right is the old gold file - they look pretty much identical (no matter how I slice the results). Also ran Exodiff on the two meshes - a handful of nodes are off by ~1e-3 or less.

As for the CIVET failure, I have no idea - nothing has changed with the test recipe and we just merged Walid's PR yesterday (which ran the test suite fine). Can we narrow it down to the MOOSE update? (e.g. could try separating out the transfer update from this PR into its own)

I'll open a new PR with just the submodule update to isolate the build issue. Also building on Pinchot at the moment to attempt to replicate the failure in an environment I have access to.

Update: I cannot replicate the build issue on Pinchot and #1272 (just the MOOSE submodule update) fails standalone.