Primary continues accepting transactions beyond FastStartFailoverLagLimit when Observer is down

Oracle 19c EE 2-node RAC with Dataguard in Max Performance Mode and FSFO enabled

During testing, there are some unexplained observations:

1. Standby Down scenario: Primary continues accepting transactions. Once FSFLL (FastStartFailoverLagLimit) has elapsed, Primary takes permission from Observer and continues accepting transactions. This is understood and clear.
2. Observer Down scenario: Primary continues accepting transactions. Once FSFLL has elapsed, we expected Primary to stall, because Observer is down and cannot give permission to Primary to continue accepting transactions, but it did not. Is this expected? Is that because AutoFailover is anyways disabled since Observer is down?
3. Observer & Standby Down scenario: Primary continues accepting transactions till FSFLL elapses, then it stalls for a period = FastStartFailoverPmyShutdown and then it shuts down. We manually bring up Standby, it starts catching up. We bring up Primary expecting that primary will stall because Standby has not yet caught up and then lag > FSFLL. But Primary does not stall instead it opens successfully and accepts transactions. (Note that Observer is down). Why is Primary allowed to accept transactions even when lag > FSFLL?