This invariant must hold if an "ordinary" within-cluster superuser is to qualify as the cluster's "bootstrap superuser":
the name of the bootstrap superuser's within-cluster role
AND
the name of the O/S user that owns lots of (but not all*) the software files that define the PostgreSQL RDBMS, together with the various files that represent what users create
are identical.
Nope, the name of the bootstrap user is the one supplied to initdb via the --username argument. Period. It need not match any name on the host operating system and it will still be the bootstrap superuser's role name.
Yes, the description for --username probably should be modified to read:
"Selects the user name of the cluster's bootstrap superuser." Or just consider a "cluster superuser" the term d'art...since most people would just refer to any old role having superuser authorization as being plain ole "superuser". The fact that is says "database superuser" is the same holdover effect as the fact that "init db" means "init database" even though it actually initializes a cluster.
Thanks again, David. And once again, all is clear now.
