> It sounds like the T production chain might itself be quite messy.

It is: it's definitely the biggest challenge after plasma confinement.

> Molten isotopes salt and lead?

There are two main blanket technology in development: ceramic and liquid breeders. They're called breeders but are very different from the kind of breeders you have in a fission reactor. Both are based on converting lithium to tritium by capturing fusion neutrons, but in one case the lithium is in the form of solid pebbles, while in the other, in a molten mixture of lithium-lead (there are no salts AFAIK).

To produce more tritium than you start with you also need a neutron multiplier: beryllium in ceramic breeders and lead in liquid breeders. The problem is beryllium is rare (and also toxic): a 500MW reactor needs ~200 kg/year, which is not a lot, but there's very very little beryllium on earth. If you factor in the initial reactor inventory (170 t/reactor) it turns out ubiquitous fusion energy it's not sustainable if we choose beryllium. If you go with lithium-lead you need more material: 3 t/year (but remember lead is a lot heavier and more common too). If you plan to cover the world energy base load with fusion, you would need a lot of lead (~10% world annual production) but it's doable.

For me, the biggest problem right now is lithium: DT fusion needs lots of pure ⁶Li, which is extracted by enriching even more natural lithium. If we're not careful enough with recycling it from old batteries, we are likely to exhaust the world resources in a few decades.

> What do you do with when it goes bad? It may not go boom Chernobyl-style, but it's still far from the birds-in-the-sky deuterium-from-the-sea fusion dream.

The worst case scenario is still the loss of coolant accident (LoCA). The blanket is exposed to a ~2MW/m² heat load from the plasma (in addition to all kind of radiation), so failing to cool adequately a module means it will very rapidly turns into a (radioactive) molten mess that's not easy to handle. Yeah, it's bad but not nearly as bad as the same accident in a fission reactor.

The bad zone of radioactive byproducts is the half life between 10 and 1000000 years. Shorter and you can wait it out. Longer and the activity is low enough to ignore when it's spread out. When designing a fusion reactor, you can usually choose components that won't generate these undesirable nuclides. But in fission, many products are generated, so persistent contamination is practically impossible to avoid.