While there are some suggesting threads …
Since I started with jME I had several issues with the updateXXX() functions. What would really help would be some flags which are reset when there is a change that makes it necessary to run one of the update functions. That way it's more unlikely that you create an overhead by calling functions too often or miss updates and get nothing on the scene. The locks are still a way to optimize and there should be ways to manipulate the dirty flags.
I remember other engines where there was a "compile scene" function which did more or less what the jME update bounds etc. does - other things were updated when necessary under the hood.
Yes, that's a thing that already is on the todo list (at least to have those flags internally and update only if needed). But as it requires some major implementation or even api changes it was postponed.