Yes, that protocol looks good to me.
I’m sorry I was unclear. What I meant with “But in this case …” was when running with couple-intramol = yes, i.e. eliminating the molecule. With couple-intramol = no the obtained DG should correspond to DG_solv, indeed, with the usual caveats for limited sampling.
There are no clear guide lines for when you can use couple-intramol = no. There is some discussion about it in the thread TI - Free Energy - couple-intramol=no - #14 by hess.