My guess is that the field index was added some time in between these runs.
Hi,
Maybe not significant, but from in the latest FME release (like since 2017 or so), the reading of a CSV file is way much faster than in older version of FME...It might be a part of the reason why it goes faster!