Parse the many-many relationship with commits and bugs

[andymeneely]

We need to handle the many-many relationship between a bug and a commit. We need:

• A new table and model to join the two
• Parse the BUG= field in the git lot parser properly - see below
• The git log loader should load up a new CSV file that gets copied to the new bug_commit table.
• Update any queries that rely on these tables
• Write dev data verifies. Add a new commit to the dev data that has multiple bugs. Find a real one in ~axmvse/chromium/realdata/chromium-gitlog.txt

For parsing the BUG= field, do the following:

1. Split along comma and space
2. Strip out the chromium: string from each field
3. If that field matches an up-to-six-digit integer, convert using to_i
4. If it doesn't match, see if the bug is a repeat copy with a regex (just like in our dev data).

[andymeneely]

Also,

• Remove extra BUG= strings
• And downcase it

And, for 4 above, use this regexp: ([0-9]{3,6})\1

then use the group from the first. Be sure to write a verify for this based on the copied one from our dev data (229611229611)

[andymeneely]

Also do this:

• Remove issue
• Remove the equals sign if it starts with the equals sign

[Felivel]

Working on verifying the reason for the dangling bugs.

[andymeneely]

The ~70k missing bugs from this relationship are key - we need to be scraping these ASAP.

[Felivel]

I placed the recovered bugs in /tmp/recovered_bugs/ they can be added to the production build data. A total of 4201 bug ids returned error [403 Forbidden, 404 Not Found], when trying to fetch the data. The error log is located at /tmp/recovered_bugs/error_log.csv