Presence spec points assume ATTACHING status is always caused by an ATTACH operation

[lawrence-forooghian]

** @(CHA-PR3d)@ @[Testable]@ If the room status is @attaching@, then the @Enter@ call will wait for the attach operation to succeed and then call the underlying presence @Enter@ operation. It will throw an error if the attach fails.

(and similar points for other places we use channel presence operations)

This assumes that a room can only be in ATTACHING because of an ATTACH operation. This is not true — it can also be in this status because:

• a contributor entered ATTACHING and the transient disconnect timeout subsequently elapsed (CHA-RL4b7), or
• the RETRY operation put it into ATTACHED (CHA-RL5f)

How should the CHA-PR3d wait behave in this situation?