Don't bother with the small delays. By the time you call the function to do the delay the 200 nanoseconds is already over. Also, the delay() function takes an integer number of milliseconds. For delays shorter than 1 millisecond, use delayMicroseconds() which takes an integer number of microseconds.
The assembler example at the end of the document is probably as close as you will get to a real specification. The main difference for the Arduino is that you probably can't treat a pin as both output and input like the assembler does. You have to use pinMode() to switch modes.