This PR adds a better implementation of the ability to find better FullMatch memories. The code is similar to the HLS implementation, where for a particular tracklet, the best values are stored, and new stubs are only matched if their delta values (z, r, phi depending on barrel or disk matches) are better than the current best.
PR validation:
We were already seeing full agreement with 100 events in the barrel. However, while developing and testing the HLS MC disk implementation, I've seen this gives more correct matches (the HLS implementation is stull under development so not all matches agree yet).
PR description:
This PR adds a better implementation of the ability to find better FullMatch memories. The code is similar to the HLS implementation, where for a particular tracklet, the best values are stored, and new stubs are only matched if their delta values (z, r, phi depending on barrel or disk matches) are better than the current best.
PR validation:
We were already seeing full agreement with 100 events in the barrel. However, while developing and testing the HLS MC disk implementation, I've seen this gives more correct matches (the HLS implementation is stull under development so not all matches agree yet).