We transfer |targetOrigin| parameter of postMessage into the target OOPIF (see RemoteDOMWindow::SchedulePostMessage and FrameMsg_PostMessage_Params::target_origin) and verify that the origins match (see how LocalDOMWindow::DispatchMessageEventWithOriginCheck is called from RenderFrameImpl::PostMessageEvent).  This security check may be skipped if the receipient's renderer has been compromised by a malicious webpage.  Therefore, we should enforce |targetOrigin| in the sender's renderer and/or in the browser process.

We should also audit whether MessagePort and other messaging APIs need a similar enforcement (MessageChannel/MessagePort doesn't seem to have an explicit |targetOrigin|, but MessagePortProvider::PostMessageToFrame still ends up populating FrameMsg_PostMessage_Params::target_origin).