I do feel that adding the indexes of the blobs and wrapping that around the PFB so we could have a compact fraud proof for a PFB that doesn’t have the blobs it paid for included was a premature optimisation.
IIUC indexes of blobs were not originally added for compact fraud proofs and instead were added when the block proposer had discretion over where a blob was placed in the square layout. The blob index was needed for validators to construct the same data square that the block proposer created. The need for the blob index was removed when square construction became deterministic which removes the block proposer’s ability to add arbitrary padding between blobs.
I agree that this would have been a premature optimization if blob indexes were strictly for compact fraud proofs. But I think the index wrapper constraint was needed for a previous iteration of the protocol design.