Comment by gpsx

It looks like they measured the time of the process where the entanglement occurred, not a time of entanglement itself. I guess they can use this to put an upper bound on the time for entanglement, which is a valid thing to do for an instantaneous process.