Adding multi-locale support, clarifying documentation, and expanding …

[capncodewash]

…tests.

This pull request does a few things:

• Adds the ability to process TMX files with more than two locales. It therefore also restructures the resulting DataFrame columns. There is no longer the concept of a 'source' and 'target' locale inside the DataFrame itself - each column is named after the locale code, and these columns contain the corresponding content. This means the DataFrame is 'narrower'. You can still get information about the source locale from the 'metadata' object.
• Clarifies the readme file (the example import statement didn't work).
• Adds an additional test to check the multi-locale support.
• Adds the ability to deal with elements containing complex content, not just plain text. This is a bit crude just now, but it allows us to deal with text containing placeholder tags and so on.

Thanks for your work in creating this very useful project, Jade!

Graeme