The actual report will continue to be piped into reports/python_complexity.log.

By exposing the paver output (approx 4 lines), we'll be able to consume
the actual complexity number in our logs and therefore trend the results
in our log aggregator.