Currently validate_drp uses some manual crude addition of calibration information and constructs new schemas to store this information. This is essentially what TransformTask is meant for. Using this would simplify the code, make it less fragile, and ideally eventually integrate more transparently with future calibration improvements or redefinitions of how zeropoints are tracked..

1. Learn how to use TransformTask. Note DM-4948 is the doc task for this.
2. Adapt the code.
3. Verify unchanged results on existing validation_data_decam and validation_data_cfht.