1. Short lifespan for unique units doesn't sound right. They need to be important for the civ.
2. Making them harder to build would increase the problem, not decrease it. It would be even viable for people to skip those unique units altogether.
3. This could be good, like ability for dedicated culture civ to still build decent units. But in big picture it doesn't change anything. You even get those units earlier than the weaker unit of the same niche, making this unit redundant, or it comes too late, making UU redundant.
1. That doesn't necessarily follow. If Samurai, say, have approximately the same strength as a musket (they don't, but for the sake of argument) but come one full era before the musket they will be a fearsome unit indeed. And instead of "short lifespan" I should have written "short
er lifespan" than a non-UU, it seems like the UUs we've seen have a decently long lifespan - plus, if a UU is strong enough, there may be some that will be useful long after they're technically obsolete (for example, ranged mounted units in Civ 5 like the Camel Archer or Keshik).
2. I'm not sure what you mean, I think it would be easier to balance a very strong UU if it must be built from scratch: If one can mass-upgrade the moment the tech is researched it stands to reason that the UU would be balanced toward the weaker end of things. If built from scratch it can be stronger.
3. Re: redundancy of regular units - possibly true, depends on what their roles are. Samurai and Pikes are both unlocked at the same tech and the Samurai is stronger than the Pike so why even include it? Because the Pike offers defense against mounted, so it can fill a different role. Re: UU redundancy - well that's only going to happen if the UU is balanced downward (ie, it's just weak and not worth getting anyway) or if for some reason you don't plan for it.
The point I am trying to make is that since UUs aren't always tied to a normal unit now the devs have the freedom to really fine-tune their abilities, costs and strength.