Add some telemetry to Syslog

[ruslandoga]

And maybe we could add some metrics as well, but I'm leaving it out of the PR since I am not seeing metrics for any other backends.

syslog_metrics = [
  distribution("logflare.syslog_pool.connect.duration",
    event_name: "logflare.syslog_pool.connect.stop",
    unit: {:native, :millisecond},
    description: "Syslog pool connection latency"
  ),
  counter("logflare.syslog_pool.connect.error",
    event_name: "logflare.syslog_pool.connect.stop",
    keep: &error_kind?/1,
    tags: [:reason],
    tag_values: &low_cardinality_syslog_reason/1,
    description: "Syslog pool initial connection errors during checkout"
  ),
  counter("logflare.syslog_pool.disconnect",
    tags: [:reason],
    tag_values: &low_cardinality_syslog_reason/1,
    description: "Syslog pool disconnections for established connections"
  ),
  counter("logflare.syslog_pool.reused_connection",
    description: "Syslog pool reused connections"
  )
]

defp error_kind?(%{kind: :error}), do: true
defp error_kind?(_metadata), do: false

defp low_cardinality_syslog_reason(%{reason: reason} = metadata) do
  low_cardinality_reason =
    case reason do
      :timeout -> "timeout"
      :closed -> "closed"
      :econnrefused -> "econnrefused"
      :nxdomain -> "nxdomain"
      _ -> "other"
    end

  %{metadata | reason: low_cardinality_reason}
end