If that is true then the ck shouldn't allow you to do something that crashes ingame, should have a size limit or something. Anyways, the bug is up there, that's all I want, to report it.
Cheers!
the CK editor works a little differently than the run time creation engine (gamebryo). fundamentally, the CK's render window does not represent a true real-time environment (which is in part why running the CK app is way more system intensive than running tesv.exe), and as such it cannot exactly predict something like this (i am not a software engineer these are just my educated guesses). this is the same for any game engine editor like Unity3D and UDK, although those 2 have an in-editor real-time preview. the CK does not.
as far as i know there are no hard measurements any of us know of to calculate exactly when that ceiling is reached. it doesnt have anything to do with physical size (total area of the mod), it is not even entirely based on a hard number object count. i am pretty sure the esp "runs out of available memory" based on how the engine calculates all of the simultaneous renderable assets, and (this is where i personally believe the bug happens) when it loads a new area, it doesn't purge the ram thoroughly before the next area starts loading in more assets (hence causing a memory crash).
for all we know maybe esp's are supposed to have the same memory ceiling as esm's, but the game engine itself mistakenly refuses to purge all of the ram before loading the new area on esp's only. this would require a low-level engine fix if this is true. because if it is true, the CK is doing its job by allowing you to create the intended number of assets per cell. the error is on the runtime engine side