You think "cum granulo salis" is evidence for G being an original part of the Latin alphabet, while not being evidence that U was an original part of the Latin alphabet?

Sure, G wasn't "originally" part of the Latin alphabet—until they invented that letter.

What do you think makes G different from J and U in that regard?

Timing. Where do you think G slots in the timeline, relative to the Latin classics?

G dates to the classical period. But, as I pointed out above, it's still intrusive enough then that Caesar's praenomen, Gaius, is written C, not G. The name is older than the letter, but younger than Latin writing.