Refactored histograms filling code for speed

This MR refactors internals of HistogramFillers so that they are stateless. Also the monitored variables to filler variables matching is redone. All in all a speedup of about 30% is achieved and we have a bit simpler code. Remaining speed optimisations seem to be achievable in fill method implementation in particular fillers. tagging @fwinkl @ponyisi @cburton