Allow for more values of windowbits

[mkoculak]

While trying to figure out why Matlab can read its own compressed .mat files that have a missing checksum, I found info that it just doesn't check it and that this can be replicated with zlib with setting the proper negative value of windowbits.
It works when I comment out the check:

CodecZlib.jl/src/decompression.jl

Lines 80 to 82 in 8ac30ef

	if !(8 ≤ windowbits ≤ 15)
	throw(ArgumentError("windowbits must be within 8..15"))
	end

but maybe we could just allow other ranges of valid values (or at least the one from -15 to -8)?

[nhz2]

To get the negative window bits, you can use DeflateDecompressor instead of ZlibDecompressor.

CodecZlib.jl/src/decompression.jl

Lines 124 to 131 in 8ac30ef

	function DeflateDecompressor(;windowbits::Integer=Z_DEFAULT_WINDOWBITS)
	if !(8 ≤ windowbits ≤ 15)
	throw(ArgumentError("windowbits must be within 8..15"))
	end
	zstream = ZStream()
	finalizer(decompress_finalizer!, zstream)
	return DeflateDecompressor(zstream, -Int(windowbits))
	end

[mkoculak]

Oh yeah, that works. Thanks!
So just to clarify things fully - because I saw in the documentation of zlib that yet another set of values of window bits gives the gzip variant - in CodecZlib these three options zlib, gzip, and deflate are just aliases to the same call to zlib with different window bits?