That all sounds correct. Plasma is more conductive than Earth's gaseous atmosphere by many orders of magnitude. I think the arc usually forms a disconnect within 10 seconds or so, because of thermal turbulence it has helped induce in the surrounding air. Problem is, you don't want a line arcing for 10 seconds, because that can really do some damage. Does 10 seconds sound about right? Or will it go on indefinitely? Of course, it's dependent on voltage, spacing, even geometry. I guess I mean in general.
From the grid side, we clear a fault like that in less than a quarter of a second. It can wreak havok when a fault lingers. But sometimes they do! There's video of a fault that lasted long enough for someone to wake up at the sound of an explosion, look out the window, grab their phone and start videoing. My experience, which mostly comes from training on "this is why we always clear faults quickly," is a fault that remains for a few seconds eventually becomes more severe. I think it's that turbulent air you mentioned. One of the three phases will have sustained plasma, and that plasma moves around and eventually the second and third phases get involved. If the first phase wasn't enough to clear the fault, the second and third ones are. Plasma is really interesting!
I'm typing this while my girlfriend watches, this is like my dirty talk. Here is what she has to say: y,k6gfhu9-ewatfrdrtqyhg6f5e478huhiiiuy Edit: we are lol'ing together. OK, now we are cuddling Edit 2 (the morning after): Uh. Thank you for your tolerance and clarification. :)