The driver event for remain-on-channel may be delayed in a way that
allows management-frame-received event to be received before
wpa_supplicant knows that the driver is actually already on the
previously requested channel. We should not request a new
remain-on-channel to send a response to just a frame if we are waiting
for the driver to get to the same channel. Instead, just continue
waiting for the driver event.
When the first wpa_supplicant interface is not the correct one for
transmitting an Action frame (e.g., P2P Presence Request frame
uses a group interface), the code got stuck in an infinite busy
loop. Fix the iteration to go through the interfaces properly.