Is your feature request related to a problem? Please describe.

The deeplabcut benchmark scripts currently do not check if a submission is complete, e.g. if predictions for all test images are returned, as noted by @n-poulsen .

Describe the solution you'd like

A good solution would be to add a few lines to this part of the evaluation code, sth along the lines of

predictions = self.get_predictions(name)
self._validate_predictions(name)
...

def _validate_predictions(self, predictions):
  # check if predictions contains too many images not contained in the ground truth
  # check if predictions contains not all images in the ground truth
  # (other potential tests)
  pass