If a node was primary in 4.3 but restarted with a 4.2 binary after an unclean shutdown. The fastcount of oplogTruncateAfterPoint could be wrong. This is related to SERVER-41391 where we added a function to correct the fastcount. But since 4.2 doesnt have that ability (unless we backport that), restarting a node with a 4.2 binary after an unclean shutdown in 4.3, the fastcount could be wrong.

We do not even support a crashed v4.3 binary being restarted with a v4.2 binary. So putting the requires_fcv_44 tag to the test.