Fix a couple of bugs when using condensediffs.
Review Request #7773 — Created Nov. 17, 2015 and submitted — Latest diff uploaded
The condensediffs command had a couple of regressions that weren't caught before release. The first is that our loop to find RawFileDiffData objects was, upon encountering an IntegrityError, attempting to look up the entry using the wrong hash. We wanted to use the hash provided on that instance, but were using a different value, leading to a DoesNotExist error. The second is that our percentage for the progress bar was overly accurate, showing the entire floating point number. It now shows an integer value.
Ran condensediffs. Saw that the percentage was correct, and that we weren't
crashing anymore upon hitting IntegrityErrors.