Why has "thin blocks" never been implemented in Core? It seems like the most obvious possible scalability improvement for transmitting new blocks, and I've seen very old posts on Bitcointalk talking as though it was expected to be done. Is the fact that it provides a constant improvement that does ...