Fix episode title notifications when there are unicode characters in it.

[p0psicles]

• PR is based on the DEVELOP branch
• Don't send big changes all at once. Split up big PRs into multiple smaller PRs that are easier to manage and review
• Read the contribution guide

This is reproducable by (manual) searching (and getting a result) for an episode which has unicode chars in the title.

Like for ex. Dragon ball super S04E13.
Protect Kaiō-shin Gowasu - Destroy Zamasu!

I tested this on Windows 7 x64.
Maybe it's not even an issue on linux like OS'.

[p0psicles]

Need to perform some tests on different os. To see if this is safe.

[OmgImAlexis]

Let me know what needs to happen and I can test on Windows 10 and OSX.

[fernandog]

Confirm it fixed:

My concerns:

1. if we should do in the above log to with "We have found season packs for {0}".format(self.show.name)
2. If we should add this in the ui.py to apply to all messages: self._messages.append(Notification(title, message, MESSAGE))

ping @ratoaq2

UPDATE:
i tried changing in ui.py:
self._messages.append(Notification(title.encode('utf-8'), message.encode('utf-8'), MESSAGE))

that but same error:
UnicodeEncodeError: 'charmap' codec can't encode character u'\u014d' in position 40: character maps to <undefined>

Error without the PR:

2017-02-06 13:18:48 DEBUG    SEARCHQUEUE-MANUAL-295068 :: [51ff22a] Traceback (most recent call last):
  File "C:/Users/Fernando/Documents/Medusa\medusa\search\queue.py", line 345, in run
    ui.notifications.message("We have found results for {0}".format(self.segment[0].pretty_name()),
  File "C:\Python27\lib\encodings\cp1252.py", line 12, in encode
    return codecs.charmap_encode(input,errors,encoding_table)
UnicodeEncodeError: 'charmap' codec can't encode character u'\u014d' in position 40: character maps to <undefined>

After:

[ratoaq2]

The same as in #2096
I dont think that's the right place to encode and decode. I haven't check the source for those strings, but the encode/decode should be done quite earlier in the flow.

[p0psicles]

This is different then the read from file imo. We should come with a better solution for this.

[p0psicles]

I was thinking about utilizing the str magic method on the series and episode objects. But then we should remove the current str mess. As I don't think it's really used anywhere.
Then we can by default encode the name using encode('utf-8').

I still have to work out some details though.