[hotfix] tag healthchecks for exclusion from prometheus message counter

context

• we count all messages written to the signald socket to a prometheus counter that we use to monitor traffic
• healtchecks (which go out in bursts of hundreds every 10 minutes) make the graphs for this counter very noisy

changes

• this introduces a conditional check in socket.write to differentiate healthchecks from other sent messages so we can exclude them from traffic graphs