Nothing wrong.
It does send the whole ~64K firmware, because your program is in reality just few functions that the main part calls in its main loop. The extra slowness is explained somewhere in this topic:
(My PC won’t load that particular topic for some reason so I can’t give direct link, but it is post ~ #40.)
With decent computer and usb, clean compile is ~15sec and flash 10sec. Second make when the core libraries are already build, the complile only takes few seconds. So if you like to make little tweaks and try them everytime the code changes even a bit (like me), it is really worth the trouble to set up local development enviroment.