[$] A RDRAND followup

In a recent episode, 'Pitchforks for RDSEED',
we learned that there was some uncertainty around whether hardware-based
random-number generators on x86 CPUs could fail. Since the consequences of
failure in some situations (confidential-computing applications in
particular) can be catastrophic, there was some concern about this prospect
and what to do about it. Since then, the situation has come a bit more
into focus, and there would appear to be an agreed-upon plan for changes to
be made to the kernel.