Cherry-pick bot should not stop on a closed branch if further pick targets remain

Cherry-pick bot currently halts when it encounters a closed branch. This was originally not an issue when all picks were being performed in parallel, but after moving to the waterfall model of cherry-picking, failure to skip over closed branch targets has resulted in a great deal of hanging changes that have failed to make it into the LTS branch targets below the closed branch.

Following internal developer discussion and feedback, this possibility of a regression to due to including a change in an auto-detected "closed" branch is so minimal that the bot should skip these detected closed target branches and continue picking. Branches should only ever report as closed when they are such. False positives for this should not be possible under the current permissions scheme.