[BUG] *TO/VO/DO crashes when no quantities tracked in cost/constraints and no Tracked Quantities folder present

[bfregly]

Describe the bug
Eliminate the need to have tracked quantities data for quantities that are not tracked in the cost function or constraints.

If a TO, VO, or DO run does not track certain quantities in the cost function or constraints, the run will crash if the Tracked Quantities folder is not present with all of the typical subfolders and files present. To get the TO, VO, or DO run to work, you have to create a bogus Tracked Quantities folder with bogus (though formatted in the expected manner) data files in it.

To Reproduce
Run TO, VO, or DO on a sample problem with no tracked quantities in the cost function or constraints and do not have a Tracked Quantities folder with expected files present.

Expected behavior
If no quantities are tracked in the cost function or constraints, then a Track Quantities folder with associated files in it should not be required.

Screenshots
If applicable, add screenshots to help explain your problem.

Desktop (please complete the following information):

• OS: [e.g. Windows, Mac M1, Mac Intel]
• OpenSim Version: [e.g. 4.4]
• MATLAB Version: [e.g. 2022b]
• NMSM Pipeline Version: [e.g. v1.0]

Additional context
Add any other context about the problem here.

[RobSalati]

When you use a ground contact model, you are required to have a ground contact file even if you are not tracking the ground reactions. Are you suggesting to make this ground reactions file optional, or is this about something else?

Currently, all files are parsed before the cost terms, so we can't check for relevant tracking cost terms without majorly overhauling our parsing.

We could potentially play around with the logic to make ground reaction parsing optional, and then check for cost term compatibility after parsing is done.

@bfregly

[bfregly]

Yes, I am suggesting that if a model has any .osimx model entity that we want to use in a TO run but the run does NOT have any data to track for that quantity (such as ground reactions for my soccer kicking example, and maybe the issue would be the same if we had muscles in a TO run with no muscle or synergy activations to track), we should still let the user predict those quantities and not make the user create a fake data file just to get TO to run.

Once we get to VO and DO, it's not a problem to predict ground reactions (and probably muscle activations as well) without tracking them. So somehow the logic works for those two tools.

Can't we use the same logic for TO?