Are you speaking to the reliability of the launch vehicle or of the satellite? The launch itself is typically not a primary cost driver of a satellite mission. Complexity may be, due to constraints of the fairings but the manufacturing and quality checks drive quite a bit of the cost
I think they mean reliability of the satellite. The JWST is going to be parked at a Lagrange point that's going to put it out of reach of most everything. At the time it was conceived, robotic servicing was even further off. A crewed mission was right out. Costs went far up in part because unlike Hubble, servicing was going to be virtually impossible.
Yes, but Starship is a prototype at the moment. And it'd require the Super Heavy to even reach orbit, which will soon begin testing [0].
Additionally, the engineering requirements for long-term life support are significantly more involved than the Dragon capsule.
There's also the testing and certification process for crewed-missions; in non-Elon time, this is likely several years, conservatively speaking [1]. (I'd love to be proven wrong, however!).
