BUG: plicRDF not working proberly across cyclic patches

When using the plicRDF (as opposed to isoAlpha) reconstruction scheme, the RDF field in cells that are point neighbours to interface cells is calculated as a distance weighted average. Here "distance" means "distance to interface centre" in nearby interface cells. This calculation goes wrong across cyclic boundaries where the interface centre should be transformed to the corresponding ghost cell position.

To reproduce, run the tutorial interIsoFoam/discInConstantFlowCyclicBCs (e.g. latest: v2406) with "reconstructionScheme plicRDF;" in fvSolution and observe in paraview the alpha.water flotsam generated as the interface crosses the cyclic boundaries. This problem is first described by @tmaric in issue #2457 (closed) . He and co-workers at TU Darmstadt deviced a solution and a patch (see issue #2457 (closed) ) but it does not work in parallel and is not optimized for performance.

In this issue we will provide a patch based on work done mainly by PhD-student David Müller from KIT and postdoc Konstantinos Missios from Roskilde University.