The new random_directio test (developed in WT-4225) could take over schema_abort's role.  sue.loverso writes:

 A general comment: this test seems more comprehensive for the schema validation than schema_abort which does not really keep track of what should be there. It only makes sure that the crashed database can recover without error. It never has turned up a problem, nor did it reproduce the original problem we were trying to repro. It does use timestamps too though. Should this test replace that one?