Quality cuts: changing the strategy

[JulienPeloton]

We currently reject alerts based on quality cuts. I'm not sure anymore this is correct, as these cuts might be relevant for some science cases but not others. For example, I recently found that we reject a lot of alerts because their PSF is extended. But that might be perfectly valid for an asteroid outburst -- or NEO.

So I propose a new strategy. We keep the cuts, but instead of rejecting alerts, we add a new column with a flag: 0 (pass) and 1 (do not pass). Then users might decide if they want to keep alerts or not based on this flag.

I also propose to reprocess all historical data that did not pass quality cuts for SSO, and push that to HBase.