[PATCH 3 of 6 V2] rebase: use scmutil.cleanupnodes (issue5606) (BC)

[Jun Wu]

Excerpts from Martin von Zweigbergk's message of 2017-07-07 23:08:29 -0700:
> On Fri, Jul 7, 2017 at 11:04 PM, Jun Wu <quark at fb.com> wrote:
> > Excerpts from Martin von Zweigbergk's message of 2017-07-07 22:59:55 -0700:
> >> I just checked that the obsmarkers are produced in deterministic order
> >> (independent of dict iteration order). However, bookmarks get moved in
> >> dict iteration order. Maybe we should make sure that's also
> >> deterministic? Can probably move the sorting you do for obsmarkers a
> >> bit earlier and reuse that sorted list of tuples.
> >
> > Good point. I guess it is not causing problems since no test tests the
> > "moving bookmarks" debug message. But "debugobsolete" does rely on marker
> > orders. Maybe we need to make the debug message exposed somewhere and fix
> > the sorting.
> 
> I now see that I wasn't very clear, but the obsmarkers already seem to
> be created in a deterministic order (by increasing source rev number).
> It's just that the bookmarks that should probably be moved in the same
> order.

I think I understand what you mean. IIUC, bookmark movement order only
affects the order of printing the debug message "moving bookmarks %r from %s
to %s". We don't have tests for that debug message yet so the undefined
order isn't causing issues (like, it does not change the end result of
.hg/bookmarks)