Support text extraction from eml files and other text extraction improvements

[martinburchell]

document_to_text() now supports .eml files, with attachments processed by any supported document converters. The exception here is we don't try to run the fallback strings converter because we would end up with a lot of rubbish from images etc.

I've had to rename the email module to avoid conflict with the Python standard library. At the same time I've renamed json and profile to avoid any future conflicts.

A few other minor changes to text extraction:

• Distinguish between missing and empty blobs
• Make the HTML parser more tolerant of dodgy encoding
• Fix warnings from BeautifulSoup

[RudolfCardinal]

Looks good - thank you! I note:

• In get_filelikeobject you've changed if not filename and not blob to if filename is None and blob is None; that does remove a check for empty strings or empty byte-like objects. Perhaps for blob that was the intention, particularly for empty e-mail content? There's another instance of this later (line 1428) too.

[martinburchell]

Looks good - thank you! I note:

* In `get_filelikeobject` you've changed `if not filename and not blob` to `if filename is None and blob is None`; that does remove a check for empty strings or empty byte-like objects. Perhaps for blob that was the intention, particularly for empty e-mail content? There's another instance of this later (line 1428) too.

Yes it was necessary for empty email content (supported by the tests) but I'll fix it so there is a better error message for an empty filename.