I recall (?) those codes are documented, somewhere, by Cisco, but Dan's post answers your first question.
Dan's second answer might confuse a tiny bit. TFTP runs over UDP, but the TFTP protocol, itself, is what determines whether data has been received or not.
Wanting to "know" whether your data has been received is a common network application requirement. Either the "application", in this case TFTP, needs to perform that function, somehow, or if you use a network protocol like TCP, TCP will perform that function for the higher level "application". (Again, this is such a common requirement, it's one of the primary reasons TCP was defined, so that each network application didn't need to reinvent that wheel. [Another common feature often desired is some form of transmission rate control, which is provided by TCP, but not by TFTP. This is why FTP generally operates so much faster over a WAN , as it use TCP, versus TFTP.])
