In any case, going forward, France's cost for new reactors are as exorbitant as in the US, if not worse. The new reactor at Flamanville has been a disaster.
At an exchange rate of 1 franc (1993) to 1 usd (1993) and 1 usd (1993) to 1.77 usd (2019), this yields a cost of 120 billion dollars. Not bad to decarbonize the electricity supply of a country with a GDP more than 1/8 that of the US.
The nature of construction nuclear plants (and many other engineering products) is that that serial production is cheaper than the first several models, since the cost of development needs to be recouped. There are only 2 EPRs in operation and 4 in construction. By comparison, during the Messemer plan, 30 reactors of the same 900 MW design and 20 reactors of the 1300 MW design were constructed. This isn't counting reactors exported to different countries. Of course first 6 EPRs are going to be more expensive per unit as compared to serial runs in the dozens.
If people are willing to pay enough for the 20% of time that nuclear stays in business. If not, we make do with just renewables.
The practical issue how how the nuclear plant sells electricity can be handled either with contracts (they could sell electricity to resellers in "one month of continuous production" types of units) or by running them in load-following mode, french/german style, whichever is more profitable.
In practice it seems fairly likely that there are many electricity applications where is pretty inelastic up to at least 10x current cost, so there will be demand for some amount nuclear power.