fix: material scan benchmarks allow_failure true

Due to a bug in ROOT, the material maps fail for the polycones in the sciglass barrel ecal. In particular, the recursive determination of DistFromInside in the ROOT polycone class geom/geom/src/TGeoPcon.cxx is leading to a stack overflow. I suspect this may be in particular due to the constant eta scan line being aligned just along the polycon eta surface of the same eta.