usbus/hid: fix ep_out readiness

[dylad]

Contribution description

This PR corrects the readyness of the OUT endpoint for HID.
Before that, EP was configured as waiting for OUT data during the initialization phase.
Unfortunately, during the enumeration phase, the USB endpoint is reset and thus, the endpoint configuration is lost.
Somehow, nRF52 and SAM0 usbdev implementation are able to workaround this issue but this should be fixed on HID side.

Testing procedure

Flash tests/usbus_hid and follow the test procedure.

Issues/PRs references

None.

[bergzand]

Ack, tested on nucleo-f746zg. Fixes the issue for me

[Ollrogge]

This PR breaks functionality of tests/sys_fido2_ctap. Readding the call of usbdev_ep_xmit to _init fixes the issue.
I don't understand the relationship between signaling to be ready for data from HOST and the SET_IDLE request. Could you explain?

[dylad]

I don't understand the relationship between signaling to be ready for data from HOST and the SET_IDLE request.

Strictly speaking, there is no relation between them. In fact, tests/usbus_hid wasn't working on ST hardware and I also had issue with an ongoing usbdev implementation I am working on.
I move the ep_xmit function here, as a workaround, to ensure OUT endpoint are ready to accept data after a USB reset. Not the best place I guess but the idea was just to be able to fix broken platform.
I was unaware of the existence of FIDO, and that it uses HID otherwise I would have give it a try before hand.
If a call to usbdev_ep_xmit() fixes tests/sys_fido2_ctap, then let revert part of this PR.
Would you mind open a PR for that ? Otherwise I can probably do it during the weekend.