If you allow the time to go to infinity, on the assumption that power continues to be produced at arbitrarily low heat output (which it doesn't), and that the conversion system maintains the same efficiency (it doesn't), then you get simply:
$$E_\infty=P_0{t_{1/2}\over\log 2}$$
where $P_0$ is the initial power and $t_{1/2}$ is the half-life. This assumes that only one decay is contributing to the energy output. It gets more complicated for a chain of daughter products, but that won't matter in this case if you only go to a few hundred years. To cut it off at time $T$, the energy up to that point is:
$$E_T=P_0{t_{1/2}\over\log 2}\left(1-e^{-{T\log 2\over t_{1/2}}}\right)$$
Since you asked about going down to some fraction of the original power, let's call that $f$, this simplifies to:
$$E_f=P_0{t_{1/2}\over\log 2}\left(1-f\right)$$
So for $f=0.1$, $t_{1/2}=87.7\,\mathrm{y}$, and $P_0=125\,\mathrm{W}$, we get $E_f=449\,\mathrm{GJ}$. That's a while though, since $f=0.1$ at $T=291\,\mathrm{y}$.
The above significantly underestimates the reduction in power of a real RTG. The degradation of the thermoelectric converters can equal or exceed the decay rate of the Plutonium. See this presentation for more details. The half-life can however be used as shown above to estimate the total heat output of the RTG, if $P_0$ is the initial heat flux.