Neat, these do explain some of the pain.
Basically, the GROWTH_CEILING caps out very quickly at 3+3, and then above 50% capacity the lower threshold starts choking even that lower number off -- and it seems like the LOWER_THRESHOLD is not scaled by the GROWTH_CEILING so you suffer a penalty which would compensate for the uncapped growth number, even though you never get that much growth to begin with.
The LOWER_THRESHOLD is just a conditional floor of 0 on the bonus growth modifier (and by default, it applies when you are below 50% capacity). You don't start getting penalized as soon as you go over 50%, it's just that the floor is gone, so the modifier *can* go negative (exactly when depends on the shape of the curve, but it certainly won't go negative at 51%). Until you hit zero, LOWER_THRESHOLD doesn't do anything.
The low GROWTH_CEILING, however, is a problem. It means Ecumenopolises and Ringworlds still make a negligible contribution to population growth compared to their capacity (they only manage 3+3, which as you say is achieved quite quickly on much smaller colonies), and so they have to be "fed" with a bunch of regular planets and habitats. This is especially bad with the empire modifier to pop cost, but even without it, it would be an issue. So I would effectively remove the ceiling, but make the gradient of the curve a bit less and/or lower the base amount of growth, so that big planets with substantial populations actually count more in proportion to their size, whereas colony spam doesn't count for so much.
Also, this is probably not possible by editing the defines files, but one for the devs: if a planet has net emigration, that should count as a certain amount of extra capacity, because the logic of "my kid won't have enough space to live" doesn't apply when you know for a fact your kid will be living on another planet with plenty of space. This sort of change is needed especially to make a plausible model of early colonization, where the homeworld is crowded but most population growth happens there, while the colonies are mostly empty and grow mostly by immigration.
Getting around that penalty via vassals feels pretty well designed - it incentivizes you to reduce micromanagement for yourself and also makes the galaxy a bit more dynamic in the late game, since your vassals might rebel against you.
Compared to other Paradox games, Stellaris is extremely pro-overlord. Theoretically vassals can rebel, if you are at peace and they have a realistic chance of beating you. But if either of those is not the case, they're doomed. Vassals do not get a veto on the integration process, even if they are "disloyal". The options for another independent empire to intervene are also pretty limited, beyond the basic "support independence". So the dynamism is pretty limited in practice.