fix: unpublish dns-sd service

[gmaclennan]

We were not unpublishing the zeroconf service when the app went into the background, but we were re-publishing the service when the app came into the foreground (the same with wifi coming off and on).

When zeroconf attempts to publish a service with the same name, it appends (${n}) to the published service name.

The result of this was that the app would leave multiple published dns-sd services e.g.

8130fb3b21ac48b6
8130fb3b21ac48b6 (1)
8130fb3b21ac48b6 (2)

Only the most recent would have the correct port - the older services would have the port from previous app opens.

Leaving these services published does not break anything per se, but the consequences are:

1. Increased mdns network traffic, because other mdns devices share the service information.
2. Slower connections between CoMapeo devices and possible performance overhead, because CoMapeo will attempt to connect to every service published.

This PR correctly unpublishes the service when the app goes into the background or the wifi turns off. Unfortunately the react-native-zeroconf library has not implemented an event for publishing or unpublishing failing, so we implement a timeout, but don't throw an error, because just trying again is a reasonable solution. We do report the error to Sentry though so that we can track these failures.

I have tested this by opening and closing the app while viewing this DNS-SD Browser for MacOS, but you could also test via avahi-browse -a -v --ignore-local --resolve.

I do not have two devices at the moment to test on real devices, this would benefit from that.

I don't think this is necessary for MVP, but could be worth releasing in a hotfix if there are discovery issues with large numbers of devices on the same network.

[gmaclennan]

This is possibly creating too much noise with Sentry because of reporting timeouts when stopping service adverstisement. Could investigate when that happens (probably when wifi is turned off) and be more selective in what we report to sentry.

[EvanHahn]

LGTM overall, but a few nitpicks.

[EvanHahn]

This is possibly creating too much noise with Sentry because of reporting timeouts when stopping service adverstisement. Could investigate when that happens (probably when wifi is turned off) and be more selective in what we report to sentry.

I didn't address this comment because this PR was blocking a release, but I still think it's worth considering.

I'll bring it up with the team soon.