I think this is one better suited for the community to test as even if it works for a given user with a given config it’s not officially supported by the SOC and in turn would be impractical from a business standpoint for purism to support. I think most of us would rather purism spent that time on further development than testing something saying “the non-supported configuration didn’t work” then all it takes is one person to say “it’s working for me” and they have to waste time with people going “why won’t you sell it this way, you must have tested it wrong, it works for that person”; or worse, they test it, it works, they say it works, then a security fix breaks whatever allowed it to work at the time of testing/it works on their test systems but not all systems and they are now dealing with support requests of "you said it works but my computer won’t boot in this config that isn’t technically supported.
Purism’s best option, in my opinion, is to do like they’ve done; share that it’s not supported and do nothing else. If the community finds a way that works for some but not others purism can simply point to the “we passed along that this isn’t officially supported, our advise is to go back to a supported configuration”.
This is something every hardware vendor I’ve dealt with does and I find it to be a reasonable approach. Support what is supported and let others experiment with what “might” work. I personally had an Intel motherboard working in a non-standard configuration for years before a bios update broke that but fixed a different issue I had, which is why I mention a security fix breaking the non-supported configuration.