In my experiments I was using ceramic tube with melting point around 2000°C but it was melted too - I was using Kanthal A1. But when the winding is failing arcs may occur. Arcs temperature can be much higher than 2000°C and can last for tens of seconds. You can see my heater after failure - to this day I dont know if it was caused by arc or excess heat. Actually both phenomenons could appear but still we dont know if excess heat from inside caused this failure.
It was so hot, that stainless steel plate that was 5cm away was melted too (by air heat transfer).
You can see small bubble at the bottom that was formed during failure. Alumina cement was rated for 1800°C.
Looks like You have not cured cement before you applyed power to coil. In this case arcing is possible but any way not melting the ceramic tube. If you can not cure heater correctly just decrease amount of cement. Nicrome heater works fine without any cement. Ufortunately FeCrAl(Kanthal A1) should be cemented. Use cement as low as possible. Another problem is green rust at negative end if you use DC. In general I'd like to say there are several issues on this way that have to be overcome. Arcing, green rust, gas tighting, hydrogenation,steel type etc. Be ready to analyze and separate them. Prepare the solution.