Revisit the GC perftests

[smlambert]

The tests in the perftest dir were running regularly on an internal server back in the days before omr was open-sourced. It would dump the results to a DB to track performance over time.

I would like to revisit these tests, reenable the test runs and monitor with our test results summary service to track GC perf at this low-level over time. This issue is to track and discuss this work.

I will also plan to update the tests to utilize Google benchmark which integrates nicely into the existing Googetest framework that is used at the project to enforce a standardized way of running/tracking benchmarks as we consider expanding out the set of low-level benchmarks that can be created within this project.

For measuring GC perf, we can (and will) still use the xml config files to populate the GC object model (based off of large/real models found in core files from actual application crashes).

[github-actions[bot]]

This issue is stale because it has been open 180 days with no activity. Remove stale label or comment on the issue or it will be closed in 60 days.

[smlambert]

It's possible we may be able to nudge this one forward. Since it was last running, there has been a fair number of changes:

• with the move to cmake, we have to ensure perftests can and are being built
• with the existence of TRSS, we can monitor test runs at ci.eclipse.org/omr (so can gather and store perf data in a TRSS DB (mongoDB backend)
• using Gtest benchmark as stated in the description is not necessary as a first step, though still an interesting exploration as it may come with some desired features
• verify that the basic structure of GC logs is the same as when these tests were initially written (or update appropriately)