MCP: Fix boundary issues in tight WP

This MR fixes some boundary issues that result in interpolation errors from TH2, reported by @wenhao. The changes are:

1. For |eta| == 2.5, reject the muon before attempting interpolation. There is no effect on physics, as the muon was rejected before and is still rejected now. However, we avoid the printing of the error message from TH2.

2. For pT == X, we choose the histogram defined for pT in [X,Z) rather than [Y,X). There is in principle an effect on physics, as the muon would be indiscriminantly rejected before and is now judged by the appropriate cut values. The impact should be extremely minor, as the change affects only muons with pT satisfying a floating-point equality.