The tiny 3mm LEDs have a thermal coefficient of about >200’C/W and 9V*120mA ~1W is not a sensation temp rise towards 200’C but if you can get several Watts thru a 5mm LED rated for 65 mW, with 9V the target temp rise is more sensational. The difference is the ESR of the LED. (V-Vf)/ESR is approx current rise. The ESR is inverse to power rating.
Try an ATX 5V power supply on a 5W white LED with an ESR from 0.1 to 0.2 Ohms and the 2V excess voltage will rise to 10 to 20A or 50 to 100W and be blinding bright and very hot very quick.
Then if MTBF is 50kh (at some reasonable temp and you know temp rise in ‘C/W estimate using 50% reduction of life for every 10’C rise and see how many seconds it lasts
Put a 10A current meter in series with an MOT primary winding and a 3.7V Li-Ion cell then watch current rise towards 10A and before it goes it accelerates , core is beginning to saturate , then pull the wire off and watch the nice long arc until it extinguishes. But no arc on contact as I rises slowly dI/dt=L/V. It’s a 1 second test.
Reverse polarity cap tests should be done outside as the fumes are toxic.
Put a 1 Ohm resistor across 9V Battery and before Battery gets too hot, fssst , say hmm smells like 1Ohm. But warning again fuming epoxy is toxic.
Then get a SPDT relay 5V or 12V and wire power in series with NC contacts to coil. It will buzz for about a few minutes before end of life and have a long wire loop in series with this buzzer and see if you can crash windows PC with the loop current noise and 10kV arc across contacts at some few hundred Hz. It may also jam all nearby cell phones.
But for practical tests just energize any tiny relay with a 1m loop of wire and crash (reset) every Arduino running some LED flash test with a cable and switch. (With extra ESD protection). The loop antenna is a good EMI immunity test for near field testing of sensors. Each switch off may cause a glitch and result in errors. Then try twisted pair with ferrite CM choke to see if it passes immunity test.