Add top level script to build datasets for trigger working point definition

This adds a lightweight version of the trigger top-level python configuration, a thinner configuration file, and updates the trigger grid submit script to work with either the older trigger configuration or this new light version.

The new python configuration mainly removes the track association we were doing for fullscan tracks, along with the need to run the track extrapolator to use these in b-tagging. The configuration file drops tracks, and only saves the inputs we need to define working points (several DL1d outputs, a few truth labels).

I also updated the script to submit trigger jobs to the grid. Now it has a "mode" switch, which lets you change the executable and the default configuration file. This seemed nicer than adding an entirely new script to change a few fields, as I might have mentioned in #13 (closed).