Expensive because it's a custom built physics lab, not a commercial power plant. Slow because it's an international project. Not just that, but it also requires lots of infrastructure to be built and entire industries to develop in multiple countries, before it can be useful. ITER is massive, but it's also just a tip of the iceberg.

>It just doesn't strike me as obvious that reducing the major radius by a few meters would have such a huge impact on cost/timelines.

It would, easily. Past a certain size, production costs rise exponentially and require one-off tech.