tauRecTools: don't use vertex-corrected clusters for MVA energy scale in the trigger

Hello,

This MR fixes the behaviour of the MVA energy scale calculation in tau triggers. In the HLT, we try not to use the vertex correction for cluster 4-momenta, as it's not really needed and can cause trouble. In the MVA tau energy scale, this was overlooked. Despite the "VertexCorrection" flag that controls the use of the vertex correction, the loop over clusters that decides which clusters are within 0.2 of the tau axis and participate in the tau pt calculation uses the vertex correction. The same goes for the tauIntermediateAxisEM variable calculation (variable currently not used in the HLT MVA TES, but we plan to use it for Run3 triggers). In the trigger, MVA tau pt is computed twice, at a calo-only stage where we don't have access to the tau vertex yet, and later at the precision step where we do have a vertex. The fact that we currently use the vertex correction creates irreproducibilities: pt is not always the same for calo-only and precision steps (clusters can jump in/out of the 0.2 cone depending whether the vertex is available), which is not desirable. We now force the tool to ignore the vertex in the trigger.

We expect small changes in the "perf" and "mediumRNN" trigger counts, but not in "idperf" because this hypo applies no selection (not even the pt cut). No change is reported by runTrigART.py -m -j4, maybe stats is insufficient.

Tagging @iriu , @gipezzul , @ademaria .

Cheers, Bertrand