After the CppSuite tests were moved to a new everygreen variant and a subset were made more stressful (WT-8640), the hs_cleanup test's cc_pages_removed statistic is extremely varied at the end of test runs.

The latest build shows this clearly in the performance plot generated from the test:

We can see that prior to the commit identified by the arrow the statistic was noisy to begin with but after WT-8640 the variance between commits has jumped significantly.

While its unclear if the stat should be generally higher or lower than the commits prior to WT-8640 it is interesting that it shows significant variance between commits and the reason for that should be investigated and understood.

Scope:

• Investigate why the statistic is so varied between runs
• Create follow up ticket to fix the behaviour if possible