@tron I did the tests in a 15 minute break in the rain taking photos of a brick wall at 0.8m distance, using the R6 with either the RF 16mm or the EF 16-35mm f/4 with adapter. Each shot was taken as a jpg and RAW, and the RAW processed using DxO PL5 and its lens correction profiles.
RF 16mm
Out of camera jpg had the bricks 1.093x larger than those processed as RAW. Interestingly, whereas the ooc jpgs are the standard 3648x5472px, the processed RAW are 3648x5981px, which is a factor of 1.093 wider. So, if the 16mm ooc jpgs are truly 16mm focal length, the focal length from RAW corresponds to 14.6mm
EF 16mm
The ooc jpgs were distorted, showing any in-camera correction isn't good, or non-existent. The processed RAW gave bricks 1.074x the size of those from the ooc jpgs from the RF 16mm. This corresponds to a minimum focal length of the zoom of 17.2mm.
So, it seems like
@neuroanatomist found with the RF 14-35mm, Canon's internal correction of the horrible barrel distortion works by cropping the extremes of the barrel before or after corrections, whereas the DxO PL5 uses a wider field. The extreme corners from DxO are mushy, but the areas on the RAW corresponding to the extreme corners of out of camera jpegs were as good if not better.
Checked it out at 22cm and 19m, and the results are the same, 9% larger field of view from RAW.
Edit: calculated the focal length directly from image size and it's ~14.4mm for RAW processed image, and ~15.8mm for out of camera jpegs.