Transfers for meshes with different variable types and overlapping domains

[lewisgross1296]

I've been thinking more about Transfers after reading the What Transfer should I use? tutorial and was curious about some best practices/comparisons between two transfers: MultiAppShapeEvaluationTransfer and MultiAppGeometricInterpolationTransfer.

I'm running a coupled simulation where my heat conduction app (call it MOOSE) uses a LINEAR LAGRAGNE temperature variable and my main app (call it OpenMC) uses a CONSTANT MONOMIAL temperature. My meshes occupy the same physical volume (same outer bound) so there's no overhanging region, though my OpenMC mesh "mirror" is more coarse.

Is it better to use a MultiAppShapeEvaluationTransfer or a MultiAppGeometricInterpolationTransfer to transfer the MOOSE temperature to the OpenMC app?

The Transfers tutorial warns about the case with a MultiAppShapeEvaluationTransfer where the receiving app has nodes outside the bounds of the source app (in which case those nodes get a zero value). If I have perfectly overlapping meshes -- albeit different discretization within, is one better than the other? Is it more accurate to query "the finite element solution from the source variable at each receiving point in the receiving application" (MultiAppShapeEvaluationTransfer)? Or is it better to transfer "the nearest node's source variable to the nearest node on the target mesh using mesh interpolation" (MultiAppGeometricInterpolationTransfer).

Since they use different shape functions, the way the MultiAppShapeEvaluationTransfer is worded sounds seemingly better? I've been using MultiAppGeometricInterpolationTransfer all along so far though (as I believe the tutorials I've been following have also used the same transfer).