Hi, PDP-8!
Interesting observations. I've checked and can approve your guess, that window positioned at 0,0 is maximizing much faster, than non-cornered one. But as i use Core on very slow box (compared with modern hardware, its lightning fast for me, my memory isn't damaged yet and keeps some reminiscences of 80386, 80286, 8086 and, yes, PDP-11, even soviet MIR-1 computer with 16K RAM cube of ferrite cores) - Atom Z520 - i must say that Mounttool behaviour is the same, faster a little but the same, i see it on my box.
So when you call maximize cornered window is spread on the whole screen immediately. And if window is uncornered, it is first moved to 0,0 corner (i see it) and then spread around. And during size restore all is taken the reverse order - squeezing to the 0,0 corner and then moving to original position.
Seems that moving window takes more time than spreading it over full screen.
I like Xfbdev too! I am TinyCore addict and as every addict don't want to be treated )
There are some issues with ctrl-alt-V and ctrl-alt-H, if you are interested i can describe them.