spawn new workers if backlog is > 0 for some time

[prymitive]

With this patch cheaper busyness algo will check if backlog is > 0 for too long, if so it will spawn additional worker(s).
This patch tries to address this case:

• user starts app with 2 workers (cheaper=2, workers=16)
• app gets to "heavy" requests that take long time to process
• until at least one of those requests is finished all new requests will be kept in backlog
• depending on how user configured busyness algo it is possible that if requests in backlog are there for too long they might get timed out

[prymitive]

This one is simple, tested (I'm running uWSGI with this patch since 2 weeks) and it does it's job properly.
It really helps when running with just few workers.

[unbit]

can you re-make the pull request ? (even if i do not get why i cannot merge it...)

[prymitive]

I was about to commit cosmetic fix ...