ENH: Enable parsing of date from filename (without strict filename structure)

[jat255]

This PR adds a new (optional) feature that will pull a date out of the filename. I have a document source that is often just YYYY-MM-DD followed by some text description, and so none of the strict date parsing from https://github.com/danielquinn/paperless/blob/5c3cb1e4ab2bb286659339c721323406c5f7cc04/src/documents/models.py#L431-L435 was working for me. We already have a great date parser for the text of a file, so this extends it to the filename as well, provided PAPERLESS_FILENAME_DATE_ORDER is defined in the config file.

I also modified the massive date regex to permit year-first dates, as well as match on dates that are followed/preceded by either - or _ (rather than just word breaks).

Against @danielquinn's wishes, I run with a time zone specified in my settings, and so I've changed the date tests to be specific to a user's defined time zone, as well as made the date formats explicit in each test, so there shouldn't be problems testing my code from Europe, or running the original tests in the US. I also added a blurb to the documentation, as well as a couple tests for the filename date parsing, and ensured that all the tests passed on my local machine after these additions.

Please feel free to add comments/suggestions/etc.

[danielquinn]

All done :-) I had to go through and conform the code to the style guide, but outside of that, it's some great work and great to have in the project, thanks!