BTaggingSelectionJsonTool: more robust fix concerning pTBins and efficiency cuts matching (following up of MR !78570)

A bug concerning the flat mass efficiency cuts was found by @caiyi (many thanks!) this has been discussed in two MMs channels

• aft-algs/channels/town-square
• aft-algs/pl

The origin of the issue is that in the previous implementation, the json utility uses a dictionnary behind the scene.
There were 2 different loops:

• loop 1: loop over pT bin values (stored in a list)
• loop 2: loop over the pT keys (pT_200_250, pT_250_300 ect) to retrieve the corresponding (mass bin, efficiency cut) values

--> The loop 2 could introduce some inconsistency w.r.t order of loop 1 because the pT key order could be modified by the json "dictionnary" utility

This MR implements a more robust fix than already provided in MR: !78570 (merged) as by modifiying the logic.
We now have a single loop where at the same time pT bins values and (mass bin, efficiency cut) values are retrived ensuring the consistency of the information retrieved

For simplicity, let's imagine that there is 3 pT bins in the JSON file e.g. pTbins : [ 200, 250, 300, "inf" ]

In the loop over the pT bins values we successively

• retrieve the pT value so succesively pT will correspond to 200 then 250 then 300 then "inf"
• for each pT value we then retrieve the corresponding pT bin name that is called pT_key
  so succesively pT_key="pT_200_250", pT_key="pT_250_300" and pT_key="pT_300_inf"
• for each pT_key
  • Make sure that the pT_key can be found in the JSON file
  • then retrieve the corresponding (mass bin, efficiency cut) values

--> In this way there is no more issues things are consistent

Tagging @dguest, @vvecchio, @ligang, @dbaronmo and @bdong