test_checkpoint currently does two consecutive searches in workers.c::worker_op. The intention is to check for non-repeatable reads, but this can be extended a little bit by also checking the key and value, not just the return from search_near/search.

As part of WT-9257 I had a quick attempt at this, but there's an existing bug on develop that I didn't want to include in my PR. The trouble is that the value will sometimes be a little different after the search call compared to after the search_near call. I didn't do a tonne of debugging on this, but it's not consistent and only seemed to appear when running test_checkpoint as part of the larger test suite.