samples: Add USB Audio to broadcast audio source

[larsgk]

Adds support for using a connected host to
broadcast audio via USB Audio.

Offload LC3 encoding to separate thread.

[larsgk]

Moved the PR out of draft to get some initial reviews. There are some glitches on the USB Audio side (periodic approx 10 secs of clean flow and audio, then 2-3 secs of missing data coming from USB) which may require updates to how the USB data flow is handled.

[larsgk]

Moving the PR to draft until I have the encoder in a separate thread and addressed the issues mentioned in comments above

[tmon-nordic]

The USB issues are related to following remark from nRF5340 Product Specification:
"If the EasyDMA transfer on the isochronous endpoint is not completed before the next SOF event, the result of the transfer is undefined."

Because lc3_encode() can clearly execute for longer than 1 ms, it is completely unacceptable to have it executing in cooperative thread context.

[larsgk]

ok - offloaded to a separate thread and everything flows smoothly now (cleaning up the PR). There are a very few occasional glitches, probably because we don't have sample rate conversion to compensate for clock drifting/mismatch - but rare enough for the sample IMO. Working on reintroducing multi stream (stereo) and the rest of the change requests.

[larsgk]

@Thalley and @tmon-nordic - it seems to be in a stable state now with the tests I made. Would you be able to re-review and possibly accept further improvements (e.g. to USB corner case handling) to come as a follow-up?

[Thalley]

Suggested change

	printk("Initialized ring buf %zu: capacity: %u\n", i,
	ring_buf_capacity_get(&(streams[i].audio_ring_buf)));
	printk("Initialized ring buf %zu: capacity: %u\n", i,
	ring_buf_capacity_get(&(streams[i].audio_ring_buf)));

[larsgk]

(last push was just commit message fix)

[Thalley]

Good enough as is, with some improvements that can be applied

[Thalley]

Rather than doing

CONFIG_USE_USB_AUDIO_INPUT=y

in samples/bluetooth/broadcast_audio_source/boards/nrf5340_audio_dk_nrf5340_cpuapp.conf, you could follow the approach above and do

default y if SOC_NRF5340_CPUAPP

So that is is applied for both the regular and audio kit

[Thalley]

Would be interesting to find a way to set CONFIG_USB_DEVICE_PRODUCT="Zephyr Broadcast Source" for any device that enables USB, and not just the nRF5340 ADK

[Thalley]

Should we return here, or allow only setting up a subset of the encoders?

[larsgk]

@tmon-nordic could you check if you could approve? GH still tells me you have a change requested. If anything, could it be a followup PR?

LC3 processing was moved to preemptible thread, resolving USB issues

[tmon-nordic]

@tmon-nordic could you check if you could approve? GH still tells me you have a change requested. If anything, could it be a followup PR?

I just dismissed the review because the USB issue is fixed. The problem was simply preventing USBD workqueue from processing events within the deadline (SOF = 1 ms). The PR looks fine to me, but I am not really the correct person to approve it.