remove improper clearing of callback state for web workflow background callback

[dhalbert]

• Fixes microcontroller.reset() causes board to hang indefinitely on subsequent restart #10957

@grgrant could you test? Thanks. You can use the artifact builds when they finish.

There were two issues here:

• background_callback_add_core() was not checking the whole list of callbacks to see if the one being added was a duplicate. It was only checking the first one. It was also checking cb->prev for NULL, for reasons that I don't believe made sense.
• The web workflow background callback was being zero'd out in supervisor/shared/workflow.c before it was added as a background callback. This meant that if it was already on the background callback list, its next and prev fields were being smashed, and that damaged the list structure

These problems only started being visible after #10840, the ESP-IDF v5.5.3 update, for unknown reasons (maybe some subtle FreeRTOS or wifi changes) and got worse with 10.2.0-rc.0, perhaps because of some workflow changes for Zephyr.

@grgrant I tested with a couple of different programs, to make it easier to wait long enough to see what was going on. One waits 15 seconds, and the other waits for a button push.

import time
import microcontroller

for i in range(15, 0, -1):
    print(i)
    time.sleep(1)

microcontroller.reset()

import time
import board
import digitalio
import microcontroller

button = digitalio.DigitalInOut(board.A0)
button.pull = digitalio.Pull.DOWN

for i in range(10000):
    print(i)
    if button.value:
        microcontroller.reset()
    time.sleep(1)

[grgrant]

Tests might be a little late for me sent DM with more info

[dhalbert]

Making some changes pointed out by an LLM code review, so draft for now.

EDIT: removed some incorrect changes in background_callback.c. The only real problem was the zeroing of .next and .prev in workflow.c. Re-tested on a Metro ESP32-S3.

[grgrant]

Four out of four boards I tested that failed under rc0 succeeded under 10966. I only found the earlier artifacts on 31aab5f I wasn't able to find artifacts for the LLM changes you made later as ed14186.

[dhalbert]

@grgrant For some reason my second push did not force a rebuild, so there were no artifacts! I pushed a comment-only commit and was able to get all the jobs to re-run. Here are the new artifacts: https://github.com/adafruit/circuitpython/actions/runs/24775236913?pr=10966

[grgrant]

All four boards I tested passed using the latest artifacts -- no bug seen.

Feather ESP32-S3 N4R2
ESP32-S2 TFT
QtPy N4R2
Unexpected Maker TinyS2

[dhalbert]

@grgrant thank you for the quick testing!

[tannewt]

Thank you! I'll be happy to move to a proper RTOS.