Features:
	Command line options (see below)
	Get LCID from Unix
	Initial header timestamp
	HHP processing into CHM
	Hard-linked files
	Options used to expose assumptions in the Microsoft implementation
	   "     "    " test robustness of decompressor implementations
	allow sequential compiling of several files
	multi-threaded compilation - create directory during lzx compression
	thread-safe compilation to allow simultaneous compiling of several files

Command line options:
	-i - input
	-o - output
	ITSF format version (2/3) - default=2 since it uses less space(8bytes)
	Window size code - automatic mode that checks the total input length & guesses at which ws to use
	Directory chunk size - plus an automatic mode that calculates an approximate optimum that doesn't waste too much space
	Quickref density - optimise somehow
	LCIDs - specify
	Bugginess of Transform/List or don't add it.
	Add directories?
	Add paths?
	Zero free space?
	Subdivide frames?
	Some zip/tar style options (-x etc)
	Storing files uncompressed
	Don't delete output on failure
	optimise order of offset for related files (faster decompression)
		eg if index.html relies on logo.png, styles.css & scripts.js, then crunch these files into the lzx near each other
	look for files that have overlapping data & have them overlap in the LZX stream
	--verbose
	--version
	--help
	--compatible
	print stats
		amount of bytes wasted
			in directory
			in unknown/unused/useless(Transform/List) fields
		compression ratio
			per file
			of lzx
			of whole file
		num files
