Backing out a backout? Merge with ancestor confusion.

[Doug Philips]

Question about proper Mercurial workflow.

Situation:
1. experimental changeset X is accidently merged into the mainline. (changeset X is on a branch, but not a named branch).
2. A few merged changesets later, '1' is discovered and a backout changeset for X is merged into the mainline.
3. Much later the review for changeset X is completed and it can be properly merged into the mainline.
4. Unfortunately Mercurial rejects the attempt to merge with X, complaining that it is a merge with ancestor.

I know why Mercurial thinks that, the question is what is the proper course of action:

a) Create a backout changeset for the backout changeset of X?
b) Make a new changeset that is a duplicate changeset  X's changes (obviously it will have a different changeset id) This seems wrong despite being expeditious.
c) ?? Something else that I didn't think of.

Thanks,
   --Doug