Hexagonal geometry specialization of MultiAppGeneralFieldUserObjectTransfer

[jortensi]

Reason

In multiple projects we have encountered transfer problems in hexagonal geometries. We have to use the fixed_bounding_box_size algorithm which adds runtime and requires a lot of trial and error to ensure that the transfers are working correctly. This leads to significant cost and delays in the analysis tasks.

Design

Recommend the adoption of a hexagonal mesh transfer with the ability to support mesh displacement.

Impact

Significant reduction in the complexity to setup and run these models since it is expected that MultiAppGeneralFieldUserObjectTransfer will be widely used in the data transfer of large conduction and fluid problems

[GiudGiud]

I could see two 'general' solutions here:

• work with groups of EEIDs. This is going be work for nodal variables vs element variables, and we ll have to add more support for nodal EIDs most likely.
• add a general local BoundingBox / Lattice concept, with a HexLattice instantiation. And variable transferred would be grouped by "Lattice cell" instead of by "nearest position"