Replacing the \s whitespace characters with normal spaces (" ") to prevent breaking on unicode whitespace characters (e.g., NBSP). Without this, if a branch name contains a unicode whitespace character that falls under \s, then the branch name will be truncated.

Adding a second blank line after the newly added test case.

Stdout is being encoded as Windows-1251 instead of UTF-8 due to passing
universal_newlines as True to subprocess.

This fix avoids encoding problems on Windows, as was done for fetch() in the previous commit. However, here it is being made more explicit and adds the same fix for the pull function.

…lures

Removing this block of code that strips away nonprintable ASCII characters, as it is causing some tests to fail inappropriately.

Successful tests are identified by the suffix ", done", which is being removed from some messages due to this aforementioned block of code. The codeblock probably intends to just strip away traililng non-ASCII characters from strings, but does not break, and so will instead continue to iterate through the string and then strip away both printable/nonprintable ASCII; this is causing the loss of the ", done" because anytime a nonprintable character is anywhere but the end of the string then the entire section of string after the character is truncated.

Looking back through commits, this codeblock was added in 882ebb1 as a workaround after the addition of universal_newlines. Seeing as we have removed universal_newlines in the previous commit, we can also remove this codeblock.

In future, if something likw this is needed, then maybe a join that concatenates all printable ASCII characters would serve better (e.g., "".join(b for b in line if b >= 32)), instead of trying to cut out the nonprintable chars.