Suggestion: better handling of `repo.set_head(pygit2.Commit)`

I have tried to emulate git checkout --detach <commit> in pygit2, to create detached HEAD checkout, and I did come up with the following rough equivalent (though the following does not uses transactions as git does):

rev = repo.revparse_single(commit_name)
repo.checkout_tree(rev)
repo.set_head(rev.id)

However when I tried to use simply repo.set_head(rev), I got the following error:

TypeError                                 Traceback (most recent call last)
<ipython-input-35-9612528ec0ac> in <module>()
----> 1 repo.set_head(rev)

/usr/lib/python2.7/dist-packages/pygit2/repository.pyc in set_head(self, target)
    293
    294         # if it's a string, then it's a reference name
--> 295         err = C.git_repository_set_head(self._repo, to_bytes(target))
    296         check_error(err)
    297

TypeError: initializer for ctype 'char *' must be a cdata pointer, not _pygit2.Commit

It would be nice if pygit2.Commit object was automatically peeled, but if it is decided to be not good idea, then what it needs is more user-friendly error message than the one above.

pygit2 0.26.0

Start at Repository.set_head in the repository.py traceback and compare the shown revparse_single, checkout_tree, and C.git_repository_set_head flow. Check existing tests or documentation for accepted target types; done means Commit input behavior is settled and covered, including a clear failure message if it remains unsupported.