- "1." is the safest, isolating the result of the merge in S3
- "2." would publish the result of the merge directly in S1
So it depends who needs the result of this merge, and for what.
- if you need to test a bit the result of that merge, while you go on developing S1, then having S3 is handy.
- but if you need, for developing S1, to have S2 devs merged in it, then scenario "2." is the more direct approach.