Properly designed switches have a hysteretic snapping action on the contactor, so that there is no position of the switch that would result in the contactor not being fully open or fully closed.

The “correct” design of this concept would not have any actual current (or maybe a very small DC current) going through the switch. Like the Shelly devices in this blog post there would be a relay that actually switches the current. The switch hardware would just be for show.

This shouldn’t happen in theory in a properly engineered and built system. The switches have arc chutes inside of them already, and the contacts are spring loaded to open and close as fast as possible to minimize. On the motor end, the maximum amount of amperage a motor pulls is called the Locked Rotor Current. You can achieve LRC by stopping a running motor with a screwdriver. But a motor also hits LRC for a brief second every time it starts from a rest state. So a UL rated motor sees maximum amperage all the time in its normal lifecycle.

Alternating current helps too. That’s why many switches have no DC rating or much lower DC rating compared to AC. A common mistake of hobbyists is to undersize a switch for a DC circuit based on its AC rating.

I don't think it would start a fire since all the materials surrounding the contact area are supposed to be fire resistant, and the arch is pretty small.

I asked an electrician about this once and they told me the arcing generates heat that travels up the conductor and into the wall, damaging the insulation and sheathing over time, and eventually exposing the bare, hot (as in warm) wire in the wall. This might introduce the wire to conditions where additional arcing (and additional heat) is likely -- for one thing you probably now have multiple bare wires in close proximity -- which is a common source of electrical fires.

Try it - it won't arc.

Also an arc with little current behind it (because of the light bulb inline with the circuit), isn't going to get very hot.