[tools] Refactor Results class into ResultsTracker class based on dicts

This is part of the refactoring needed to implement a feature that allows
re-running benchmarks until needed confidence is reached.

R=machenbach@chromium.org

No-Try: true
No-Tree-Checks: true
Bug: chromium:880724
Change-Id: I45f584a3503ecf567f4c2661a302a74fc5e516af
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1581605
Commit-Queue: Sergiy Belozorov <sergiyb@chromium.org>
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#61037}