Fix BLE adv timeout; ESP: don't sched handler on adv finish

[dhalbert]

Fixes #9359.

• Correct handling of advertising timeout value in both espressif and nordic common_hal_bleio_adapter_start_advertising()
  • Timeout of zero means wait forever. Use BLE_HS_FOREVER for espressif. Use BLE_GAP_ADV_TIMEOUT_GENERAL_UNLIMITED for nordic.
  • Handle zero timeout in the base routine _common_hal_bleio_adapter_start_advertising(), not in the parent routine common_hal_bleio_adapter_start_advertising(), because callers to the former expect zero to work as an unlimited timeout.
• Espressif: remove background_callback_add_core(&bleio_background_callback() from _advertising_event(). Scheduling the background task caused advertising to start up again. This caused a tight loop of advertising starting, finishing, and restarting, slowing down CircuitPython greatly. The reschedule does not appear to be necessary: it is not done in the equivalent nordic code. There is other code to restart the BLE workflow advertising as needed.

Tested with heart rate monitor, iBBQ, and BLE UART.

@bablokb Could you test to see if this fixes #9359 for you? Thanks. It might also prevent #9362, if that error is a side effect of the tight loop mentioned above.

(Note: I have been unsuccessful in getting BLE workflow to work with PyLeap with 9.0.5 or recent builds including this, but I might be be doing something wrong.)

[bill88t]

ESP32 feels a lot snappier with this!

[tannewt]

Thanks! Did you try File Glider?

[dhalbert]

Thanks! Did you try File Glider?

I tried that too and didn't get it to pair. But I need to read the Guide again.

[bablokb]

Tested with the code mentioned in #9359:

show_colors (sleep): 3.001
show_colors (title): 0.0569992
show_colors (ui): 0.0790024
show_colors (refresh): 0.392006
show_colors (run): 0.485001

load_image (sleep): 3.003
load_image (title): 0.0540009
load_image (load): 0.107002
load_image (refresh): 1.17
load_image (run): 1.289

So we are back to normal. Thanks for tracking this down.
Regarding #9362: this is not fixed yet. I will add additional infos in that issue.