Refactorings + Bugfixes

[dbrgn]

Reorganized imports

• Added future imports
• Correct grouping of imports

Moved common code from environment subclasses to CommonEnvironment

• get_os_string is already in CommonEnvironment and does not need to be redefined.
• Moved is_file method to CommonEnvironment.
• Converted methods of environment classes from staticmethod to classmethod.
• Added or corrected many docstrings.

Encoding Fixes

• Fixed many encoding issues (the software previously crashed with a UnicodeDecodeError on my Debian system). Please take a look at the remarks in the commit message: 41e277cebfc506e6f910334b2a07e06db2c0596b

Other Fixes

• Fixed dangerous kwarg default. Never use mutable objects as kwarg default.

Please review, @cfaessler

[dbrgn]

I also added a test for the common environment. It currently fails. We still need to decide how to handle valid/invalid files. Is everything except a directory a valid file? Or is only a file (type f) a valid file? Or maybe files and symlinks?

Btw, this PR raises landscape rating from 83% to 89% :)

[cfaessler]

Nice work @dbrgn, thx! This whole thing is becoming an engineering masterpiece!

[cfaessler]

[dbrgn]

Before merging, we still need to decide whether we count symlinks as files.

RIght now the algorithm for is_file is as follows:

1. If the path doesn't start with `/`, return False
2. If path cannot be opened with os.stat, return False
3. If path is a directory, return False
4. Otherwise, return True

If that's fine, I'll adjust the docstrings and tests accordingly. I'm not sure whether we should allow "is_file" to return on symlinks, named pipes or other strange file-like objects.

Possible object types, according to man find:

• Block device
• Character device
• Directory
• Named pipe
• Regular file
• Symlink
• Socket
• Door (Solaris) <--- WTF

My suggestion would be to include regular files and symlinks.

RFC, @d22 @cfaessler