Fixes the connection between primary and secondary particles

[jdkio]

Properly connects secondary particles to the parent particles in the caf, see issue #116. Also adds validation script.

# Before fix
=== validation summary ===
file: default/example-CAF.root
entries checked: 13
secondaries checked: 10426
ancestor refs resolved: 0
ancestor chain matches where CAF chain is complete: 0
errors: 10426
warnings: 0
FAIL

# After fix:
python scripts/validate_caf_truth.py fixed_secondaries/example-CAF.root 
=== validation summary ===
file: fixed_secondaries/example-CAF.root
entries checked: 13
secondaries checked: 10426
ancestor refs resolved: 10426
ancestor chain matches where CAF chain is complete: 10426
errors: 0
warnings: 0
PASS

The file sizes are not that different

3.5M    default/example-CAF.root
3.5M    fixed_secondaries/example-CAF.root

Run times are not either (note these were measured on two different days):

Original: 3m31s
Fixed: 3m29s

Note about intermediate secondaries

Secondaries are only saved if a reconstructed track warrants it. This makes sense so we're not save too much info. But this meant that some intermediate particles weren't saved. Like if a secondary neutron creates a proton track, which we reconstruct, then there was no parent to the proton in the caf. Below is the result of the validation if we don't save the interemediate particles. The fix was to save the intermediate particles. It's a more general and stable solution than trying to link the secondary to the primary. Currently the unreco'd particles are not individually flagged but we could add that if need be.

python scripts/validate_caf_truth.py particle_fix/example-CAF.root 
=== validation summary ===
file: particle_fix/example-CAF.root
entries checked: 13
secondaries checked: 7680
ancestor refs resolved: 7680
ancestor chain matches where CAF chain is complete: 5279
ancestor refs not independently checkable because CAF parent is missing: 2401
trace_missing-parent: 2401
errors: 0
warnings: 0
PASS

Files touched

src/truth/FillTruth.cxx
.gitignore

Files added

scripts/validate_caf_truth.py

[chenel]

This is a great implementation, and I'm especially happy to see the validation script---that'll be a big win in the future if/when we have to check whether future changes break anything. I'm happy to approve.

There are a few nits recorded below, with one possible edge-case bug that I'd like your further opinion on.

[chenel]

Okay---sounds good. In that case, if you have the check already instrumented, it might even be worth contemplating whether we should make it an abort, so we know to go figure out what happened?