Well, air only breaks down at an electric stress of about a million volts or so per metre - so that explains the small flash for the last quarter mm on AC and when making contact, but you might ask why does a 240V arc draw out for about an inch on trying to switch off?
At the instant of breaking the circuit, the gap is very small, fraction of a mm, and the 240V will jump that - and as it does so, that spark, if and only if there is enough current available, creates an arc- a region of gas where the electrons and the nuclei of the atoms are separated into a hot plasma - which is actually quite a good conductor - and effectively fills the gap, as if the electrodes had been larger by the size of the plasma ball. Now, if the electrode moves another fraction of a mm, the process repeats - and the plasma ball grows, so long as the current stays on to feed it - without the current flowing, the plasma is not kept in its excited state, and atoms reform back to a neutral insulating gas.
With 50Hz AC, the current falls to zero at the reversals of the AC, and this is 'off' for long enough for a modest arc like that to extinguish.
If you have ever used a buzz-box welder, you will realise that a 50Hz arc of a few tens of amps can be stretched to perhaps quarter of an inch -DC welders are quite different.
At higher frequencies, the plasma does not have time to recombine - this demo is bursts of a few hundred KHz modulated at a audio frequencies and a good fraction of a megavolt peak, and the arc lengths that can be drawn with that sort of kit is scary.. I've not seen one this big in the flesh but even playing with a little one is pretty hair raising.
playing fast and loose with higher frequencies
Its very clear the plasma track is far from a straight line
To an extent the volume of plasma is related to the kVA available, and how gently you draw it.
Jets of air, spinning magnets and cold surfaces all conspire to quench - all of these are found in different designs of circuit breaker.
Same phenomena with a neon tube, makes it have near infinite resistance below the strike voltage, and once struck a much lower resistance , hence the need for a ballast to current limit things , - and it won't go out until a much lower voltage than it originally needed to strike it.
-------------------------
regards Mike