Wouldn’t the knowledge gained by fixing a large one-off reactor be of limited transferable value, as compared to expensive knowledge that could be used to upgrade an existing “fleet” plus improve subsequent versions?
In both cases they took a long time to come to light and the country is now stuck dealing with them across a large fraction of its power production.
This idea that mass production creates more mistakes really doesn't make sense when you actually think about it.
Cars/guns/airplanes/even furniture all benefit from mass production to keep costs and mistakes down. There's no reason why reactors would be magically exempt from it.
What you are asserting isn't the reality of mass production as experienced by industrial society so far.
1. A flaw is found in a standardized design impacting hundreds of units. Engineers identity a fix, and technicians receive training on how to apply the fix. They apply the fix to hundreds of units in a standardized and repeatable fashion. If the flaw is catastrophic, chances are this will prevent hundreds of catastrophes.
2. You have an industry producing hundreds of bespoke designs. Each time a flaw is found, engineers work to devise a solution, and technicians repair the flaw on an ad-hoc basis. The industry is competitive, so potential learning surrounding individual flaws does not propagate quickly. Each catastrophic flaw is a new catastrophe, endangering many lives.
Funny thing about auto makers manufacturing 100s of millions of units: they're still hugely profitable and despite the risk of mass recalls, bespoke designs are practically unheard of.
Now we get the argument that yes, that will happen but it will still be profitable to do so.
All of this contradicts 70 years of nuclear industry experience. It is an unsupported assertion, not a truth, that modular manufacturing will experience a beneficial industrial learning curve that will bend costs toward being competitive with alternatives.