Logspout drops the first few lines if nothing has been logged for a while

We recently upgraded to Logspout `v3.2` from `v2`, and many things are better.

We have it running on a machine which runs jobs on cron twice a day. We noticed that the log entries for these jobs, which are usually only a few lines long, were truncated from the top or missing altogether when we used `v3.2`. Subsequent jobs, if run within a few minutes, are not missing any initial lines.

From https://github.com/gliderlabs/logspout/issues/309#issuecomment-306818766 I see that logspout is expected to drop logs while it is reconnecting, and we do see the `reconnect successful` message immediately before the truncated logs.

Is there anything we can do to mitigate this?

Start by reproducing the twice-daily cron-job pattern with a long idle period, then inspect the reconnect behavior described in issue #309 and the “reconnect successful” message. Done means determining a mitigation or fix that preserves the initial Docker log lines after an idle reconnect, with regression coverage for the truncated and missing-log cases.

Written by the indexing model from the issue text.