Made easy support for providers with flaky/unreliable checkip servers

[troglobit]

In #305 @themiron came up with the brilliant idea to simplify using another checkip server for providers with a flaky/unreliable service.

1. Update documentation with helpful hints and recommendations
2. Add syntax sugar a-la checkip-server=default to tell inadyn to use its built-in default (ipify) instead of the provider provided one
3. Possibly also detect bad checkip servers and warn in syslog?

[RMerl]

For 3., maybe see if throttled ipcheck servers return something else than a 200 or a 30x HTTP code, and in which case act accordingly (fallback to default server perhaps?)

[troglobit]

@RMerl ah, great idea! I was thinking about checking for timeout, but just tracking the HTTP return codes (maybe with a counter and threshold?) is quite a bit more obvious :)

[RMerl]

@themiron, did you get a particular HTTP response code when you were throttled in your test case?

[themiron]

For 2 - currently get_address_backend() uses cmd, iface and remote in order, seems easy to add last resort with default ip checker. after that, same default server can be removed from plugins for simplicity

[themiron]

@themiron, did you get a particular HTTP response code when you were throttled in your test case?

ARAIR yes, something either 4xxx or 5xx, not really sure.

[RMerl]

So, code >=400 could be handled as a failure then.

[troglobit]

Agree to all above! Thank you guys for the awesome feedback :)

[themiron]

So, code >=400 could be handled as a failure then.

already is