The X11644A and P11644A cal kits are almost identical and share the same manual. But when I look at the TRL isolation definition for the X11644A, it has no standards assigned. In contrast the P11644A has offset load, fixed load and sliding loads. Here's the X11644A definition, taken from a N5247A PNA-X.
and here's the P11644A definition, taken from the same VNA.
Another entry that appears odd to me is the H/W ratio of the waveguide.
For WR90, the width is 0.9" and the height 0.4". so H/W=0.4444444444.. However, this is rounded to only 3 significant figures (0.444) in the definition. Given the physical dimensions of the waveguide given in the X11644A manual are 22.86 mm x 10.16 mm with a tolerance of +/- 0.01 mm, it seems odd to maintain the waveguide to 4 significant figures, but give their ratio in the cal kit definition to only 3 significant figures. (I guess this is better than what I see someone do recently, which was to round |S11| in dB to 1 significant digit, then perform calculations based on that and quote the results to 6 significant digits!)
Are there a couple of minor errors in the X11644A definition, which could be improved in a later firmware release? Or is the problem with the P11644A file? I notice that the example file "Example file with unsexed connectors" has nothing for the isolation in TRL either.
FWIW, the last time I looked, the X11644A file for the 8720 was still seriously flawed, that makes it impossible to do a single port calibration on port 1 unless the file is modified. I guess there's less inclination for Keysight to sort out an 8720 issue than there is a PNA-X issue, although the X11644A is still a current product, so it does not seem unreasonable the files for it are corrected.