One more bit of info if it helps.
I noticed that I had some ProLiant DL360 Gen9 servers that booted fine with that version.
The main difference between the ones that failed booting had more CPU cores detected than the ones that succeeded.
For reference, the successful one had 20, the failed one had 32.
Given the kernel panic seems to be scheduler related, are there any chances this is an edge case for servers with tons of CPUs?
Cheers,