Independent track analysis for pure 3D tracks with XYZ hits

[jgalan]

Recently there was some issues with XYZ tracks reported by @DavidDiezIb , and addressed also in #39.

It is a bit strange for me that in the following code lines

tracklib/src/TRestTrackAnalysisProcess.cxx

Line 972 in 6b3c09c

auto hitsBoth = {hitsXZ, hitsYZ};

the XZ and YZ hits are combined to produce XYZ observables.

Indeed, this is likely producing XYZ observables out of XZ and YZ hits. However, the pure XYZ hits might require dedicated attention. Also, the TrackAnalysis process is becoming too wide.

I think we should favor smaller and more atomic processes that target a reduced problem.

We could for instance have TRestTrack3DAnalysisProcess that will target only pure XYZ tracks produced with pixel readouts or pure MC simulation output. This process would serve only for those events that contain pure XYZ tracks, so there should be some error output in case we find other track types inside the input event.

Also, we may think to split into TRestTrackBasicAnalysisProcess and TRestTrack2DAnalysisProcess, etc.

I think atomic processes should be favored.

[DavidDiezIb]

TRestTrack3DAnalysisProcess implemented here: #43
Close related with Issue #41