Let's say the number S19J ASICs sufficient to keep the facility heated even on the peak (i.e., coldest and windiest) expected day is five devices. What do you do in the fall, spring, and summer? Power down the number of them that aren't needed? Or will the excess heat just get vented out?

Unless the cost per kWh is relatively low (e.g., $0.10 per kWh or below), it may not be profitable to run the ASICs that aren't needed for the heat. And with the seasonal nature, it may not make sense to invest in the hardware if it is not going to be running more than a few months per year.

Now if natural gas prices were to rise and remain at an elevated level, then having ASICs even running just half the year might make economic sense. But there are many things we just don't know about the future (e.g., future BTC/USD exchange rate, future difficulty, future price per kWH, future price per Ccf for gas, resale value of S19J ASIC several years from now, etc.), that such an endeavor equates to speculation on a number of variable.