Split save-code-now loaders in 2 different pools
The non-git origins like svn are usually taking a long time to be processed.
In case of a campaign of submission like recently happened for sdss, all the scn worker pool was blocked on long (or never ending) tasks.
Meanwhile, the others numerous git task (most of the usual scn submissions) were stacking up.
The idea is to dedicate a part of the scn workers to git only tasks so the majority of the submissions will never be blocked by other long running scn requests.
The rest of the pool can have the same configuration so the global processing power for the save-code-now will not be affected.
The proposal is is to split the 16 worker pool in:
- save-code-now-git: 6 workers dedicated to git origins only
- save-code-now: 10 workers with the same configuration as today