`RSA12` should not be subordinate to `RSA7`, or it should be removed

From me (internal Slack message):

I notice that RSA12 is subordinate to RSA7 in the spec.. Is that a 2015-vintage mistake, or does it convey meaning (i.e. intentional)? https://github.com/ably/docs/pull/48

@SimonWoolf responded with:

It also looks unnecessary? Like, RSA7b lays out when the clientId should not be null, why do we need another spec item that lays when when it should be null? If RSA12 is the exact reverse of RSA7b it's unnecessary; if not it's contradictory (It'd also be nice if the subitems of RSA7b were explicit about their relationship to each other. Eg "If a clientId is provided in the clientOptions, or else if ...". Like... you can guess from context that they're OR'd together not ANDed, but you shouldn't need to guess from context, and it's unclear which takes priority if more than one apply.)

┆Issue is synchronized with this Jira Task by Unito

ably / specification

`RSA12` should not be subordinate to `RSA7`, or it should be removed #21