Pdfplumber

Changed
- Upgrade `pdfminer.six` requirement to `==20200104`
- Upgrade `pillow` requirement `>=7.0.0`
- Remove Python 2.7 and 3.4 from `tox` tests

Fixed
- Fix sorting bug in `page.extract_table()`
- Fix support for password-protected PDFs (PR [138](https://github.com/jsvine/pdfplumber/pull/138))

Fixed
- Fixed PDF object resolution for rotation (PR [136](https://github.com/jsvine/pdfplumber/pull/136))

Added
- `cdecimal` support for Python 2
- Support for password-protected PDFs

Added
- Caching for `.decimalize()` method

Changed
- Upgrade to `pdfminer.six==20181108`
- Make whitespace checking more robust (PR [88](https://github.com/jsvine/pdfplumber/pull/88))

Fixed
- Fix issue [75](https://github.com/jsvine/pdfplumber/issues/75) (`.to_image()` custom arguments)
- Fix issue raised in PR [77](https://github.com/jsvine/pdfplumber/pull/77) (PDFObjRef resolution), and general class of problems
- Fix issue [90](https://github.com/jsvine/pdfplumber/issues/90), and general class of problems, by explicitly typecasting each kind of PDF Object

Fixed
- Fix bug in which, when calling get_page_image(...), the alpha channel could make the whole page black out.