///// Third phase of output
//
// This is the heart of this program. It aims to be able to produce a
// sustained output rate of 64 bytes of FizzBuzz per four clock cycles
// in its main loop (with frequent breaks to do I/O, and rare breaks
// to do more expensive calculations).
//
// The third phase operates primarily using a bytecode interpreter; it
// generates a program in "FizzBuzz bytecode", for which each byte of
// bytecode generates one byte of output. The bytecode language is
// designed so that it can be interpreted using SIMD instructions; 32
// bytes of bytecode can be loaded from memory, interpreted, and have
// its output stored back into memory using just four machine
// instructions.
// [The eighth byte of LINENO_MID] changes in meaning over the course of
// the program. It does indeed represent the billions digit most of
// the time; but when the line number is getting close to a multiple
// of 10 billion, the billions and hundred-millions digits will always
// be the same as each other (either both 9s or both 0s). When this
// happens, the format changes: the hundred-millions digit of
// LINENO_MID represents *both* the hundred-millions and billions
// digits of the line number, and the top byte then represents the
// ten-billions digit. Because incrementing a number causes a row of
// consecutive 9s to either stay untouched, or all roll over to 0s at
// once, this effectively lets us do maths on more than 8 digits,
// meaning that the normal arithmetic code within the main loop can
// handle the ten-billions digit in addition to the digits below.
