Jointcal reports the collected_*_refStars and the selected_*_refStars, which should represent the total number of available refStars from the input refcat, and the number that were associated to fittedStars. The latter is the number that are important for the actual fit. Thanks to Russell Owen's investigation about a separate issue, I realized that the selected_*_refStars metrics are incorrect: jointcal does not alter the associations.refStarList during selection, but rather the pointers between fittedStars and refStars.

To fix this, we need to traverse the fittedStarList and count the number of fittedStars that have a non-nullptr refStar. Once done, we'll have to update all of the selected_*_refStars metrics.

Fixing this will help debug the jointcal test failures in DM-10765.