Then you need to try to plate solve an image. Then you will see the command that it tried down at the bottom of the Align Tab in Ekos. Based on your log, your last command that didn't work was this:
Of course that exact command is now probably invalid because the image is gone now, but basically you should try to find that complete command verbatim and copy and paste it from the log, or drag it from the Ekos window into the terminal. Then hit enter and see what happens.
I took one of my pics and used that to test this. It works but it takes a LONG time to resolve. I'll play with the command line options to see if I can speed it up and then will use that in Ekos.
The question remains how to instruct Ekos to use this PYTHONPATH?
EDIT: I timed resolving that picture at it takes 11 minutes
EDIT2: I managed to bring it down to under a minute
Ok good, Yes, you need to make sure you have the right index files and then play with the options. Ekos already uses that PythonPath. The two export commands that I told you to do are the ones I already tell the solver to use in Ekos. The only reason I told you to do it in the terminal was so that we could see more details in the output. Sorry I forgot to tell you to do the second export command in my first post. But if it worked in the terminal then your installation of KStars should work just fine for plate solving, there is no problem. Just play with the index files and the options.
Thanks again for your never failing patience. I finally managed to get it working and fast too! First of all I made sure I only have the necessary FITS files from Astrometry.net which saves the field solver from looking in too many files in case it cannot find a solution. Then I realised that since recently I am using a filed flattener that also is a 0.8x reducer. So the focal length of the telescope was wrong and thus the FOV calculated by Ekos as well. Once I solved that, all went back to nromal. Thanks once more and sorry for all the trouble.