> Such constraints were tackled by old innovations like assembling space stations using modules launched separately, or more recently, with inflatable modules like BEAM (sadly not in development anymore).
Another good example is JWST, which required an elaborate (and therefore risky) "unfolding" process. The costs of such approaches seem somewhat self-reinforcing: a failure would be very costly, so it's worth spending more on validation and testing; that extra expense would make a failure even more costly, justifying even more spending on testing! (In that sense it's similar to the tyranny of the rocket equation: having to carry more propellant in order to propel that extra propellant!)
Another good example is JWST, which required an elaborate (and therefore risky) "unfolding" process. The costs of such approaches seem somewhat self-reinforcing: a failure would be very costly, so it's worth spending more on validation and testing; that extra expense would make a failure even more costly, justifying even more spending on testing! (In that sense it's similar to the tyranny of the rocket equation: having to carry more propellant in order to propel that extra propellant!)