Basically what the title says.
My current view: The Cube LUT appears to be a bulky, uncompressed text file with little more to offer than what is can be encoded in a lossless PNG. That the main difference is the Cube's 17 times larger size (based on a single comparison with 64 dimensions), and that the Cube is more computationally demanding to parse than the PNG.
In addition, the PNG is a visual representation which gives the user an impression of the LUT even without importing it into an editor.
Why I think I'm wrong: The sheer number of LUT formats out there suggests that they offer more than what a PNG, or a combination of a PNG and a metadata file, can provide.
What am I missing?