The information regarding phase encode direction isn’t actually shown in the results of mrinfo, so it’s difficult to know for sure exactly how your acquisition protocol is designed. It does however look like you’ve got unusual diffusion encoding in that second image series: a b=0 volume, a b=700 volume, and another b=0 volume. I would advise checking the appearance of the images themselves to confirm whether or not this is actually the case.
Assuming that this is the case, and assuming that the AP encoding direction applies to the first image series (and the second series was acquired with PA direction), you would want to do the following:
The dwipreproc script is ‘hard-wired’ to deal with ‘typical’ acquisitions, to make pre-processing more convenient in those cases, partly by removing the necessity to manually define the configuration and index files for topup / eddy; but some understanding of what the script is doing, and what topup / eddy are doing, is still required, particularly in cases like this where your acquisition doesn’t perfectly conform to the expected use case.
In the particular case of the -rpe_pair option, a pair of b=0 image volumes (or image series) are provided in order to estimate the inhomogeneity field using topup; and this is used in correcting the distortions of the input DWI series, which must have the same phase-encode direction as the first of the two images in the reversed phase-encode pair, with eddy.
Those are the diffusion gradient directions for your two image series; the phase encode direction(s) is a completely separate piece of information. When I say it’s 'not shown in the results of mrinfo', what I actually mean is that the phase encode direction information is not currently extracted from the DICOM header, and is therefore not present in the header of any subsequent images. So it’s necessary to manually ensure that image pre-processing is correctly informed of this phase encoding information.
Normally this would not be too difficult as the acquisition protocol would be designed from the outset to perform pre-processing in a particular way; but given the unusual nature of the diffusion gradient table in your second image series, I would recommend checking the raw data, and/or confirming this information with whoever is responsible for design / acquisition of this particular protocol.