Thou shalt have no other gods before the ANSI C standard 1646

That was a distant secondary consideration if it was one at all. There was a general inclination to use different algorithms than AT&T did if it was better or convenient to do so, in order to avoid appearances of copying, but that's as far as it went. The PDP-11 Unix programs did contorted things to run in the 11's address space and GNU wisely avoided all that.

No, when the 32-bit policy was announced, home PC's like the original 68000-based Macintosh and the Atari ST series already had 32-bit cpu's and it was clear that's how things were going. They had no MMU's so there was a plan for GNU to do task switching by core shuffling. But the 68000 was first announced in 1979, years before GNU started, the original SUN workstations came out not much later, and that was the clbutt of processors GNU aimed at from the very beginning (the National 16032-32016 was another intended early target). The feeling was that IBM had stuffed up by using the 16-bit 8088 and the GNU developers just didn't care about it.

Also slightly misleading; GNU was aiming at somewhat different kernel choices but work was also converging to build a complete free system based on GNU stuff plus the free parts of 4.4BSD. There were several free BSD-based 386 kernels that were running before Linux 0.11 was released but attention then shifted to Linux because of the USL vs UC lawsuit. That meant the first fully free distros (e.g. Yggdrasil) chose Linux rather than BSD in order to avoid legal headaches. Linux also (my opinion) attracted more volunteer improvements the early days because it used the GNU license and was international, so it had more of a community spirit.

No, the guideline was 32 bits and 1 MB and suitable consumer machines as mentioned above were already in lots of people's homes.

Ok, good point, I forgot about that.