Not a problem.
Hmm… At first glance, this seems to be a problem with the engine itself, and how the TObjectIterator works.
For a scene with 2 actors, “1M_Cube” (a resized static mesh) and “BP2” (a blueprint containing a resized cube, like in your example), it finds these results when testing for actors with static mesh components:
Apparently, it returns every separate component (going by the C in the name) as a full AActor, and it seems sometimes the BP’s static mesh component is discovered last and returns the non-scaled size, which would lead to the wrong lightmap size application? But only if the blueprint has been added in the current session, and not after the level has been saved and reopened. Then, the only ones it found were BP2_C_0 and BP2_161, which returned the right sizes. (Well, _C_0 didn’t, but the 0 is irrelevant and anyway this didn’t appear the next time the level was loaded after an editor restart, so…)
I’m adding some code to ignore anything with C in the name as a probable blueprint component for 1.1, but does the scaling problem still happen after reopening the level after restarting the editor?
Yes, it does.
The current implementation is that it basically checks the name of the final “leaf” of the directory tree for preset names.
So you can have \interior objects\couches<presetname>, and those will be using <presetname> instead of the currently selected “main” preset.