The problem with Stirling engines is they have moving parts. It's highly unlikely a Stirling cycle generator on the Voyagers would have operated for 40+ years as their RTGs have.
The RTGs on both probes have decayed A fair amount at this point and are producing a lot less power.
10 years may seem short, but combined with an electrically powered thruster there is potential for doing types of missions we have not really been able to do before. That 10 years could be spent doing propulsion.
If you want to use it more in the science phase, use chemical rockets to get up to speed and then boot up the reactor in time to say, decelerate into orbit and you are looking at having the majority of that 10 years used at the destination.
My main concern isn't really the duration, but the reliability of the moving parts. But without plutonium, there are not a whole lot of other options for powering missions to Uranus and Neptune.
There must be some way to directly turn radioactivity (or heat) into thrust with sufficiently high exhaust velocity!?
Still hard to imagine
Not that I'm disagreeing with you really - mechanical devices have high failure rates. 10 years maintenance-free is a bit of a dream. As a mechanical engineer I'm interested to see whether the Stirling engines involved have radically different scaling to optimise for reliability, or NASA just plan to do the engineering really well.
Of course manned maintenance may be possible - they are touting this technology for Mars bases. Plus robotic maintenance is going to become more of a thing over time.
Note that while implementations usually show a crank and rod with a flywheel, it could just as easily use a magnet and coil to generate electricity.
That get's you down to a single part.
Then you have this:
https://en.wikipedia.org/wiki/Thermoacoustic_heat_engine
...that gets you down to something that can generate sound from a heat differential, and you could couple that sound to some kind of transducer to generate electricity. Still a moving part, though.
You probably can't get zero moving parts and yet have it do useful work, but you can get really close I think.
As a heuristic, the less mechanical parts in any system, the more efficient it is over time.
Ideally, everything that's built is engineered like a spacecraft, because to some extent it is, and is onboard one.
https://en.wikipedia.org/wiki/Optoelectric_nuclear_battery
Its major, still undressed downsides are the requirement of expensive beta emitters, and synthetic diamond PV cells (everything else will die to beta particles)
You get around 15 years of useful work and 15% efficiency with current day technology.
It has no moving parts as RTG, has better power to weight than RTG, and has efficiency comparable to simple Stirling