fix [home|away]_turnovers

[gt6796c]

This addresses a discrepancy between the game.xxx_turnovers in the game table and the incoming JSON from nfl.com
I don't know what 'to' is, but I'm fairly certain that it is not turnovers as the data doesn't match up with the play-by-play data. The changes here make the statistical category match up with the play-by-play.

Unfortunately, the existing data is already wrong since these values are written to the db. The following code fixes up the DB.

with nfldb.Tx(db) as cursor:
  cursor.execute("SET TIME ZONE 'UTC'")
  for year in range(start, end+1):
      games = nflgame.games(year)
      for game in games:
        dbg = nfldb.Game.from_id(db,game.eid)
        dbg.home_turnovers = int(game.data['home']['stats']['team']['trnovr'])
        dbg.away_turnovers = int(game.data['away']['stats']['team']['trnovr'])
        dbg._save(cursor)

[ochawkeye]

I think that is actually reflecting the number of timeouts that a team has remaining at the end of the game.

in 2009081551, Chicago and Buffalo both called 2 first half timeouts. In the second half, Chicago called 3 timeouts and Buffalo called 1 timeout.

[ochawkeye]

Thanks for stripping all of my comments new github (has to be their fault and not me not knowing how to use their system!)

http://www.nfl.com/gamecenter/2009081551/2009/PRE1/bears@bills#menu=gameinfo%7CcontentId%3A09000d5d811ed31d&tab=analyze&analyze=playbyplay

Chicago threw 4 interceptions and lost 1 fumble but ['away']['to'] == 0.
Buffalo threw 1 interception and lost 2 fumbles but ['home']['to'] == 2.

[gt6796c]

That was my thought as well. 2015101103 was what I was looking at. ATL should have had 3 turnovers. Didn't find something that told me how many timeouts they had left so I could confirm what exactly 'to' was. Ended up doing a search in the DB and all turnovers were between 0 and 3. That seemed.... unlikely.