I believe the SIDD images are those that had additional processing and look similar enough to GEC (unlike the more raw SICD images). In that case, our tools should work if you put the RPC coefficients into an .rpb file as GDAL expects. So, if you have a file called image.nitf, you should have a file called image.rpb. See the GDAL doc for a note on .rpb:
https://gdal.org/en/stable/drivers/raster/gtiff.html.
You can check that the rpb coefficients are read correctly by doing gdalinfo image.nitf, and see if they get printed.
We found SAR images to be very hard to process with ASP, especially when they look visually dissimilar. It is suggested to start with very similar images first. One can also reduce them to lower-resolution, to decrease noise and increase similarity, at least for initial processing, with a command such as gdal_translate -r average -outsize 25% 25%.
Apart from the issue of images being correlated well, there is the issue of how good the camera models are. We did not test SICD enough to tell.
Let us know how it goes. This will be tricky to pull off.