A fix like this is likely to rebreak in the same way. The question becomes how many times do you think it's fixable before thermal stress destroys the board. Maybe future incidents will be prevented by increasing cooling of the device in the installed system?

I'd be ok with this in a machine with regular backups, where you're looking at worst case the data from a day or two is lost, plus some downtime while getting a replacement and restoring a backup, and best case another trip through the hot air station. Seems pretty ok to me in that use. Would probably be fine for a use case with easily replaceable data too: if you've got a fast network connection, it could be your steam drive.

Yeah Steam drive is what I was thinking with large applications, but it could as easily be something like Xilinx’s Vivado…