Patchers now have access to a parse_patch_output() command, which

attempts to parse out useful information from raw patch output. This

collects patched files, files with conflicts, any fatal errors found,

and flags for whether empty files or partially-applied files may have

been present.
Patcher subclasses can make use of this in response to their own patch

tool, or even override the parsing behavior.
Much of this logic existed in apply_single_patch(), and is now called

from there instead, simplifying the function and making for a good

example for other patchers to mirror.

Unit tests passed.