It's an incredibly misleading statement, so it's not you.
Gravitational waves propagate at the speed of light, so their detection by Earth-bound detectors is expected to correlate with the arrival of light from distant events assuming the source of light generation is identical (not spatially or temporally separated) to the source of the gravitational disturbance.
In the case of a supernova, it's actually a dynamic process instead of a flip of a switch, and so the change in the magnitude of light emission can indeed lag behind by several hours from the start of collapse of the star's core - the detection of gravitational waves could allow us to "buy back" that several hour window by detecting the gravitational waves produced by core collapse instead of having to wait for the light magnitude increase. There's no disconnect here, just sloppy reporting.
In many cases however, we infer gravitational events or influences have occurred or exist by witnessing a change in motion of light emitting (or reflecting) objects that are directly affected by the event/influence - think of a supermassive black hole at a galactic center that we can't observe directly, but infer its existence by the motion of stars in its vicinity. Or the orbital behavior of Neptune that suggested other massive objects yet to be found in our solar system.
Depending on the nature of the event, we may have to infer that a black hole merger, for example, has happened by observing the changes in motion of objects we can see with traditional telescopes. This introduces a time-lag in addition to the normal speed-of-light timelag we're bound by whenever we look up at the night sky:
Gravitational influence must travel at the speed of light from the site of the event to the light-emitting object that we can observe, and then the light from that object must travel to our telescopes, again at the speed of light. At the moment that the event happened, the light from the object we're observing with our telescopes had not yet felt the disturbance, so there's an additional lag in detection time that must be accounted for - we're not really observing the black hole in this example, we're observing a surrogate object.
The ability to detect gravitational waves may allow us to "buy back" this additional lag by now 'directly' observing the inciting events... bound by the speed of light, of course.