Optimize ECAL lin-reg

[tvami]

Is your feature request related to a problem? Please describe.

This is a follow-up to #1407
There are several things to make the lin-reg better. List below.

Describe the solution you'd like

1. Implement that we really are selecting hits from the closest hits
   

   ldmx-sw/Ecal/src/Ecal/EcalVetoProcessor.cxx

   Line 996 in e6c6935

   // int hitsInRegion[50];

   
   and then
   

   ldmx-sw/Ecal/src/Ecal/EcalVetoProcessor.cxx

   Line 1028 in e6c6935

   if (trackingHitList.size() < 3) break;

   
   it should prob take the hits from trackingHitList[hitsInRegion[jHit]] and also trackingHitList[hitsInRegion[kHit]] and propagate this below in the code too.
2. That opens up immediately that 50 in int hitsInRegion[50]; might be not realistic
3. And then the distance in

   ldmx-sw/Ecal/src/Ecal/EcalVetoProcessor.cxx

   Line 1018 in e6c6935

   if (dstToHit <= 2 * cellWidth) {

   was not optimized at all.

Besides the physics changes discussed above we could also consider to move away from the ROOT approach and use Eigen instead.

Describe alternatives you've considered

Keep it as it is, currently it's not really used, but I think it has potential.