I do not think that the main problem is a focus offset between viewing and taking lens. Look at picture #2: center is (more-or-less) sharp, periphery is soft; focus offset would produce uniformly soft image of this flat and distant building.
I vote for one element having been flipped front-back. I saw this on a Fujica GS645W bought at a photo exchange, IQ was terrible; after proper re-assembly it is now delivering superb IQ.
I suggest dis-assembling the taking lens, carefully recording the orientation of elements as found. Compare with Tessar diagram. The viewing lens canot be taken as a secure reference: whoever messed up the taking lens might have put his hands on the viewing lesn also...