Introduce a factory for calo-endcap

This MR introduces our own factory for the endcap of the calorimeter. Previous one (intrinsic DD4hep factory for polyhedra endcap) was too simplistic, even compared to the respective barrel factory (which for now is quite all right).

The new factory is based on the DD4hep's one, but divides further the polyhedron in a cross section, adding a structure of modules, with a minimal gap between those modules (set to 2x2.5mm between each 2 modules). It also allows to create a hierarchy like the one present in the barrel (module-layer-slice-segmentation). Since orientation of the new modules is different than in barrel (where layers are placed along radius, while for the endcap along z=beam axis), the segmentation is changed from CartesianXY to CartesianXZ.

Here is a picture of a cross section of a single module of the new endcap, which is repeated 16 times to form a full 2-pi coverage. Note that the cells size is in reality 10x smaller.

The last commit updates the material reference ROOT file, to see a fair comparison of what has changed please consult the previous pipeline: https://gitlab.cern.ch/azaborow/OpenDataDetector/-/jobs/38476254

The change in the material does not come from the change in the factory (which recreates the same detector), but from the addition of the small air gap in between the modules, most prominent when looked at the x0_vs_phi scans.