Create macro-benchmark on fixed data

[sourcefrog]

Perhaps use http://sun.aei.polsl.pl/~sdeor/index.php?page=silesia which has samples of various file types.

Scenarios:

1. initial backup of everything
2. incremental backup with no changes
3. incremental backup with files touched but no changes
4. incremental backup with a few files changed -- open question how to generate this on these corpuses, but perhaps it could be achieved by truncating some of the files or adding others

[wscott]

Here is a benchmarking setup for backups that tries to support multiple tools for comparisons already. https://github.com/gilbertchen/benchmarking Written by the Duplicacy author. Duplicacy has source, but it not free. It has a number of interesting ideas.