Wucherl Yoo of LBL observed a scenario in which an application that caused Darshan to enter "condensed" mode because too many files were opened. In this case Darshan collapses all file records into a unified record.

The resulting log in this case, however, had several incorrect fields (including notably the start and end timestamps), and furthermore it wasn't clear from the parser output that a condensed mode was being shown.

This should be straightforward to replicate in the regression test harness.