Catch alternating errors on MacOS/BSD

[AivanF]

On MacOS (and other BSDs), when the local network is down ping emits a ping: sendto: <reason> line on stderr immediately, then the matching Request timeout for icmp_seq N once it gives up — one pair per failed packet. The timeout line is already accounted for as a lost packet, but each sendto line falls through to the generic "other line" handler and clutters the output:

• the 1:1 interleaving with Request timeout breaks the existing "Last message repeated N times" dedup,
• different sendto reasons ("Network is down", "No route to host", "Host is down", ...) count as distinct lines and each gets printed at least once,
• every such line triggers other_line_is_printed(), which moves the cursor past the stats and disrupts the live dots layout.

This adds an early branch in the awk main loop that silently drops lines matching ^ping: send(to|msg): /, mirroring the existing filters for blank lines and httping's error shutting down ssl noise. Packet-loss accounting is unchanged: the matching Request timeout still increments the counters and prints !.

Linux's iputils ping does not emit ping: sendto: / sendmsg: prefixed lines during the steady-state ping loop, so this is a no-op there. mockping.sh has been extended with an interleaved sendto/timeout block plus a recovery ping to exercise the path.