… filename is not specified.

Manual addition of some bugfixes found in upstream issues and pull
requests.

Almost certainly broken.

Numerous small changes. Update for python3, attempt to remove gzip2 and
use standard library instead. Began creating small tool for HTTP
parsing. Probobly utterly broken.

Added warcscrape.py and supporting files.

gzip.open() expects a string as filename. In the prior implementation
it passed a fileobject as argument. This resulted in a type error.
Changing it to pass the name of the file as argument to gzip.open()
fixes this problem.

  The old approach assumed to much about the internals of requests and ended up with an
  empty body, it also mixed byte encoded content with regular python strings (the latter
  was the intention so I adapted the code to that).

…aw bytes instead.

  This simplifies the code and fixes an alignment bug. It also moves the decoding responsibility
  the the user to the user of the library.