document default compression level in -h output

fix regression in 0.0.5 that would cause bogus sizes to be generated …

…for large input files

implement heuristic determination of a nice balanced dictionary size …

…providing near optimal compression at reasonably low memory usage - elzma_get_dict_size()

update ChangeLog

fix bug in serializing lzip footer, fix win32 compiler warnings

bump micro version to 2 - will move tag

port to win32