rtsp-server: If media with eos-shutdown=true is still waiting for EOS to propagate, new client will receive "Service Unavailable"
E.g.
0:00:13.742991668 1147 0xd05d88 INFO rtspmedia rtsp-media.c:3151:set_target_state: set target state to NULL for media 0x714221b8
0:00:13.743036335 1147 0xd05d88 DEBUG rtspmedia rtsp-media.c:2791:gst_rtsp_media_set_status: setting new status to 1
0:00:13.743071002 1147 0xd05d88 DEBUG rtspmedia rtsp-media.c:4065:default_unprepare: sending EOS for shutdown
[...]
0:00:32.033201670 1147 0x6f018288 INFO rtspclient rtsp-client.c:3979:handle_request: client 0xccfce8: received a request DESCRIBE rtsp://192.168.0.1/test 1.0
0:00:32.034636337 1147 0x6f018288 INFO rtspmediafactory rtsp-media-factory.c:1452:gst_rtsp_media_factory_construct: constructed media 0x714221b8 for url /test
0:00:32.036173004 1147 0x6f018288 WARN rtspmedia rtsp-media.c:3937:gst_rtsp_media_prepare: media 0x714221b8 was not unprepared
Here it took >20s for the media to shut down, which is a bug in itself. But even under normal conditions there is a race condition here between sending the EOS event, receiving the EOS message, and new clients connecting in the meantime.