This line;
------------------------------------------------------------
shows that the tail end of the log has started, and everything after that should be considered one entry.
On its own I do not think it would be that difficult to set the Logstash input multiline codec to detect this, the problem is that I already have that codec configured to detect tracebacks (indented lines) and bundle those as well. So I am not sure how to tell the multiline code to detect two different patterns. I looked into Filebeat's multiline handling, and could not figure out how to specify multiple multiline patterns on there either.
For context, this is because my log is actually the output of two different programs, the Toil CWL workflow runner, running inside of an LSF HPC job.