Do you have any idea what would nbe a good configuration for ping watchdog?

Technically the correct way to do it is pointing to some reference ip, this can be your isp router or the main AP ...

The interval is a maximum of 300 seconds for each send sequence, which means how long a request for availability of the selected IP will be sent ...

The delay is maximum 300 seconds between each reception sequence, this means that every few seconds will be required without response from the host IP, so that a fault is considered ...

And the counting fault is programmed in units, that is, every few consecutive faults must be detected before the computer restarts ....

Before this, it is trial and error, since you determine the necessary values ​​according to your scenario ...

Do not forget the restart time, so that it is correctly, the delay must be much greater than the device restart time, since it will enter an infinite restart loop ...

Finally in a point-to-point where there are two teams, usually the IPs are pointed from one to another, in a multi-Point Point scenario, it gets complicated, and generally the stations point towards the AP ...

Regards

I think maybe your typology has some problem. What about A is configured as AP mode, B and C are respectively configured as Client mode. TP-link outdoor AP is not good for bridge.