I have seen plenty of people using Arduinos and Raspberry Pis as flight computers for these applications, so clearly power consumption isn't *that* great of a concern. ;-)
I agree that an MSP430 part would likely have lower power consumption than a Hercules part, but I would be willing to bet that a Hercules part -- when used/coded properly -- would consume *considerably* less power than what most people are actually using.
I mention this application because I have seen a fair number of discussions on flight computer failures. Now personally, I suspect the primary cause to be related to vibration on spring-loaded connectors. I've seen some pretty janky wiring jobs. But cosmic radiation levels are much greater at high altitudes so this is a real possibility. (BTW, janky *is* a real word now. http://time.com/3724601/oxford-dictionary-janky-egot-ridesharing/)
I am aware that FRAM has some level of immunity to the effects of cosmic rays, but as I understand it the SRAM and the core logic could certainly be effected. As such, the effects of a cosmic-ray induced bit flip could easily go unnoticed causing a series of cascading failures before the watchdog is triggered. But then you still are left with all the mess between the cosmic ray event and the watchdog trigger. (There is also no guarantee that a failure will result in triggering of the watchdog.)
As best as I can tell, the lock-step operation of the Hercules parts would provide much quicker notification of an event, and thus much quicker restart with less chance of data loss or corruption. This is certainly no substitute for a fully rad-hard system, but it seems as if it gets you a long way in that direction for a fraction of the price.
In any case, it seems as if this could be a prime hobbyist application for the part. I am *very* curious to know if anyone who knows more about the architecture of the Hercules can actually provide some more detail here.