Match headers to group datafiles

[jordandsullivan]

Is your feature request related to a problem? Please describe.
Currently we append lines to existing datafiles if the date and hour matches up, but this can lead to issues, say if you ran upgraded a package and reran something in the same hour.

Describe the solution you'd like
Ideally in addition to checking the date/hour, we should make sure the headers match, specifically with the versions and operating system which will be added info in #144 .

Describe alternatives you've considered
We could just not append data at all to existing files, but the way our parallelism works, that would mean a separate csv file for every single combination of a compiler and benchmark file, which seems very inefficient.

Additional context