IMHO a 'heater only' solution would make a pretty poor job of taming the microbolometer as it would rely upon passive cooling in an overshoot situation.
Unlike a OCXO there is no insulation around the module to help slow Delta T effects
Why would it overshoot? It's in a fairly stable envrionment, in a metal frame moderately insulated from the outer case, so isn't going to see any sudden temperature load change.
The camera would also be inoperable above an ambient of +30C as all temperature stabilisation would cease.
Would it be inoperable, or just less accurate, or need more frequent shutter calibrations? It produces perfectly useable images from cold before it's heated up.
I don't know much about the physics of TECs, but after the microbolometer, ADC, vacuum chamber and window, there doesn't seem to be a lot of space in there for a TEC as well.
However as that sensor is also used for automotive applications, these would be specced over a wider temp range, though in that application you don't really care about absolute temp accuracy.