Populate keystone with revocation events prior to benchmarking

[lbragstad]

In previous performance runs (like with devstack), we've found that the number of revocation events stored as a significant impact on performance. We should make this a formal part of the test by adding a bunch of revocation events to the deployment before running the benchmarks.