[test] Py3 porting

[nyov]

Some py3 changes.
Fixed this only enough to make scrapylib tests pass for me.
In particular I'm not sure if this should import to_native_str from scrapy for urlpathjoin.

Can't run the test-suite locally, so someone else with access to the remote scrapinghub endpoints please take over.

[nyov]

@bertinatto, I gathered that. Would be a bit heavy for a single utility function (or three). But it was the quickest way to get the tests to pass :P
Hope the PR may serve as inspiration to someone else to do better.

This work only happened to get scrapylib tests to pass for now: scrapinghub/scrapylib#71 (comment)

[chekunkov]

IMO we shouldn't use scrapy

right

anyways, thanks for the contribution @nyov. Do you want to continue working on this PR or we can use it as a base for further Python 3 porting?

[nyov]

Please do use it as a base.. I cannot run the tests locally it seems (as I wrote earlier), so that would be too much shooting in the dark for me :)
Thanks!

(But don't thank me for the half-assed PR. It's more to raise awareness that this is a blocker for scrapylib py3 tests.)

[bertinatto]

Closing PR because #62 was merged. Thank you, @nyov!

[nyov]

Awesome. Thanks for that.