There are currently several test cases in test_cursor13.py and test_cursor16.py that have been relaxed or commented out because they erroneously reported errors as a result of the fixes for WT-6274 and WT-6463. (Some of these cases are discussed in WT-6274.)

These tests perform various cursor operations and use WT statistics to verify that WT performed the expected number of operations.  As a result of WT-6274, eviction threads now open and close history store cursors.  These operations are getting included in the statistics seen by the tests, and as a result the statistics don't match the values expected by the test programs.

There are two different ways we might fix these tests:

1. Instead of testing that statistics exactly match the expected number of operations, test that they reflect at least the expected number of operations.
2. Subtract out operations on the history store cursor from the global statistics so that we (hopefully) get the correct number of operations.  This will require first fixing WT-6478, which is incorrectly counting some per-data-source cache statistics.

Alternatively, we can remove tests that can't compensate for background operations or that no longer serve a useful purpose.