ieee802154/at86rf2xx: port to Radio HAL

[Stopkaa]

Contribution Description

This PR is a takeover of #19015 to port the at86rf2xx driver to new radio HAL. It has been rebased to the newest master with the new submac implementation from #21578 and tested successfully with the tests included in #19015.

I also encountered that with the new radio HAL, the transmission time increased by approximately 50-80ms.

Old radio HAL ping test:

2025-08-20 15:43:50,575 # 
2025-08-20 15:43:50,578 # --- fe80::204:2519:1801:bd0e PING statistics ---
2025-08-20 15:43:50,584 # 10000 packets transmitted, 9965 packets received, 0% packet loss
2025-08-20 15:43:50,589 # round-trip min/avg/max = 129.134/148.823/202.367 ms

New radio HAL ping test:

2025-08-20 16:39:35,091 # 
2025-08-20 16:39:35,095 # --- fe80::204:2519:1801:bd0e PING statistics ---
2025-08-20 16:39:35,101 # 10000 packets transmitted, 9842 packets received, 1% packet loss
2025-08-20 16:39:35,105 # round-trip min/avg/max = 182.657/220.738/2399.448 ms

Testing Procedure

Refer to #19015.

[riot-ci]

Murdock results

:x: FAILED

027a2969963e256e527e253685c64d8b81075e17 fixup! ieee802154/at86rf2xx: add initial Radio HAL support

Success	Failures	Total	Runtime
476	0	9547	02m:01s

Artifacts

[benpicco]

Great that you took that over! But why does the round-trip-time get so much worse? Since at86rf2xx is a 'full-mac' device, I'd expect the submac to do very little here :confused:

[Stopkaa]

But why does the round-trip-time get so much worse?

Actually in a debugging session with @jia200x we discovered that reverting the following commit git revert a953ae9d9fe fixes the issue

Edit from crasbe: This is the commit in question: https://github.com/RIOT-OS/RIOT/commit/a953ae9d9fe025ab6988ee0a086453bcdb7f2547