The calexp exposures in validation_data_decam have TPV distortion terms in HDU 0, which should have been stripped (and probably are when using the current master of the DM stack). In addition they have the expected SIP distortion terms in HDU 1. Thus it is possible that some FITS readers will read the WCS as having both TPV and SIP distortion terms.
Please reprocess the calexp (and all other processed files, for consistency). The current DM stack properly strips the TPV terms when generating calexp exposures, so that should take care of the problem.