In SERVER-50611, we modified this test to allow for multiple priority takeovers to occur in the test (while we only expect/induce one priority takeover to occur, we relaxed the assertion from exactly one occurs => at least one occurs, as slow elections might lead to multiple takeovers happening; see the linked ticket).

But the relaxed assertion is backwards; it asserts that the expected # of primary takeovers is >= the actual number, when in fact the actual number may be higher due to slow elections. This assertion should be reversed so actual # takeovers >= expected #.