We no longer attempt to fetch original content for new files or patched

content for deleted files. Instead, these are immediately returned as

None.
File content fetching is also better protected, handling HTTP 404 and

500 results, which both will return None. Previously, HTTP 404 results

were handled when fetching file paths, but not file content, which was

backwards. HTTP 500 wasn't handled at all, instead causing crashes.

Relevant unit tests pass.