Fix forced 30x resolution in https://github.com/jkehler/unshortenit/issues/26, add minor logging and exceptions

[fake-name]

Principally, this fixes #26.

It also adds some simple logging (at the path "Main.LinkUnshortener"), and also imports the two exceptions into the __init__.py file. This makes it much nicer to use, since you no longer need two separate imports if you want to catch exceptions.

You can now do:

import unshortenit

try:
    itemurl = unshortenit.UnshortenIt().unshorten(itemurl)
except (unshortenit.NotFound, unshortenit.UnshortenFailed):
	itemurl = None

[fake-name]

Note that I generally have fairly verbose logging, and it might be worth replacing all the log.info calls with log.debug, if you prefer the library to be more silent.

[fake-name]

Additional stupid junk: You can now provide a dict-like object to the UnshortenIt() instance, and it will then use that to allow you to provide already-resolved urls.

This was done for my use case where I have some inputs that consume RSS feeds, and the RSS feeds contain shortened links.

Due to the nature of RSS, I see the same posts fairly repeatedly, and waiting for them to resolve again, and again, and again..... This way, I can provide a custom dict-like object that saves the seen URLs in a database.

The end result is a feed where I previously took > 5 minutes to process the contained links now takes ~5 seconds.

I don't know why the pypy test is failing. It works in my local pypy3 install. Are the tests accidentally using pypy2?