Signal condition on USB error

[johanhedin]

Signal the same condition on USB error as on a successful transfer so that the consumer_threadproc can exit properly when a USB error occur (like a device disconnect).

Mitigates issue #74.

[bvernoux]

This code shall be reviewed more carefully (and tested)

• I see a potential issue if the transfer is interrupted and so device->streaming = false because of USB device is disconnected when we call airspy_stop_rx(dev) especially related to kill_io_threads() which will not clean stuff correctly because device->streaming is false ... to be checked carefully with the "full path with errors" when issue appears (especially when removing USB device during streaming)
• It seems in the code you have modified device->streaming = false shall be replaced by device->stop_requested = true for a clean stop to be confirmed/checked

[johanhedin]

• It seems in the code you have modified device->streaming = false shall be replaced by device->stop_requested = true for a clean stop to be confirmed/checked

Replacing device->streaming = false with device->stop_requested = true in the three locations my patch touches seem to work as well. But, this will remove the way to detect that something has happened since airspy_is_streaming() will now continue to return "streaming" even after the device is pulled...

And the streaming has stopped, so it feels logical that airspy_is_streaming() should return false in this case. But this can of course be fixed by extending airspy_is_streaming() to also check for device->stop_requested and return false if it is set to true.

[johanhedin]

Created a new pull request according to your suggestion (#76). Works as intended om my test setup. I will remove the irrelevant pull requests when the dust settles.

[bvernoux]

This PR is cancelled in favor to https://github.com/airspy/airspyone_host/pull/76