A single M7.9 quake lasts for a minute or two, so two quakes separated by 3 seconds would not involve two big quakes.

They may be referring to difficulties figuring out the focal mechanism (what is the orientation of the fault and which way did it slip) if the mechanism changes during the rupture. One way to get the focal mechanism is to measure whether the first motion is up or down at many seismometers, another way is to look at the long-period motions of the entire seismograms (centroid moment tensor solution). Sometimes the first motion mechanism is different from the moment tensor, leading to comments like the one that you cite.

Another possibility is that the entry of a double event was just an error. If several sources report the same earthquake with slightly different locations and times, it sometimes results in too many quakes on the earthquake lists. This is likely for yesterday's quake because the initial reports I saw had 3 events, an M7.5, M7.5, and M7.7, which within 15 minutes was corrected to a single M7.5 event, then eventually upgraded to M7.9.