Bug fix - Range checks on calib constant retriever class

As pointed out in ATLASG-2369, some of the muons were returning a NAN pt after calibration. As pointed out by @ssnyder, this was due to muon having an eta beyond the input range of the histograms. I have added a range check, and if the muon has a larger eta, the recommendation from the last bins are used (as it was intended originally)

While this is labelled as bug fix, it won't affect the current ongoing analysis as these muons will be cut away by the medium working point. The loose WP, in which these muons are include, is currently not recommended for any analysis