delayMicroseconds() is broken in AVR core versions that do LTO...

Recent versions of the Arduino IDE/Core for AVR do Link Time Optimization, which means that calls to delayMicroseconds() where the argument is constant are likely to be optimized to the point where they become significantly inaccurate for small numbers...

The simplest solution is probably to define delayMicroseconds() with a "never inline" attribute, but it would probably be a better idea to use libc and/or compiler built-ins instead.

test sketch:
```
void setup() {
}

void loop() {
byte x;
x = PORTB;
delayMicroseconds(1);
x = PORTB;
delayMicroseconds(2);
x = PORTB;
delayMicroseconds(3);
x = PORTB;
delayMicroseconds(100);
x = PORTB;
delayMicroseconds(2334);
x = PORTB;
delayMicroseconds(x);
x = PORTB;
}
```

Start by reproducing the supplied test sketch with an AVR core build that uses Link Time Optimization, focusing on delayMicroseconds() calls with constant and variable arguments. Compare the timing for the listed delays, then inspect the delayMicroseconds() implementation and compiler behavior. Done means small constant delays remain accurate while the existing sketch behavior is preserved.

Written by the indexing model from the issue text.