Add hit dumper for trigger studies

I'd like to expand our hit studies to trigger-land, specifically focusing on low-pt low-CPU reconstruction. This is the dumpster counterpart to atlas/athena!61984 (merged), which adds a flag to dump all the pixel hits into AODs.

I did a few other things with the hit writer, some required for this to work, some less so.

• Make the hit container name configurable.
• Add options to make the hit content independent of the primary vertex. The idea would be to run this before tracking, so no primary vertex allowed!
• Make all the hits variables optional.
• Add an option to change the deta variable to be hit - jet like most of the other variables. It was jet - hit. I could just change this and drop the option, but I wanted to make sure nothing changed. @sargyrop do you have an opinion here?
• Reintroduce sensible lepton overlap removal, other validation variables. Right now the trigger_hits.json config is very bare-bones.
• Get a sensible target Z position (#86 (closed)).

Since I changed the hit writer class quite a bit I tested the old vs mod h5 outputs with h5diff -c -d 0.0000001 old.h5 mod.h5 hits and found no differences.

I made some cool plots of the hits projected along the jet axis to where they intersect with the beamline. Here's the "side view" where the jet is in the image plane:

and this is the top down image, the jet is heading out of the page and to the right.