Synthetic Benchmark Giving Inconsistent Results

A licensee has noted that the synth benchmark has been giving inconsistent results when they try to find optimal scalability settings for their application. The root cause of this is unclear, but logs attached to this ticket demonstrate the issue.

1. Open a template project in the editor
2. Run SynthBenchmark in the editor multiple times
3. Note the inconsistent values on consecutive runs (for example,e FillOnly)