Fix unordered streams parsing

[hexadite-virnik]

When parsing streams, we can't detect them by thier index, because the order not consistent. we should use sections' names only.

[snemes]

This change will unfortunately break compatiblity with some files (e.g. some Luminosity Link malware samples), see issue https://github.com/crackinglandia/pype32/issues/9.

According to this blog post the metadata stream can either have the name "#~" or "#-" so we can probably replace the index based part (that was originally intended to fix it) with an added check for "#-": https://codingwithspike.wordpress.com/2012/09/01/building-a-net-disassembler-part-4-reading-the-metadata-tables-in-the-stream/

E.g. you just need to replace if name == "#~": with if name in ("#~", "#-"): on lines 1412 and 1434.