Gitlab experiment status

[Paul Gofman]

I didn't feel much benefits in the process of submitting patches through 
gitlab compared to sending them to the ML, nor any complications on that 
way. The thing which I probably find more convenient is accessing sent 
but not committed patches (sent by others): it could be I was missing 
some obvious way in the ML workflow but in general fetching a patch from 
gitlab to apply locally (using git) feels easier to me, especially if 
that involves some rebasing.