[PATCH 0 of 3 STABLE] fix some problems around un-finished histedit/rebase state

[FUJIWARA Katsunori]

This patch series fixes some problems around un-finished
histedit/rebase state.

Recent (2.7 or later) Mercurial can avoid occurrence itself of
situations focused in this patch series, because it disallows "hg
update", "hg strip" and so on while histedit or rebase is in progress.

But recent Mercurial should handle also such inconsistent
histedit/rebase state correctly, because:

  - inconsistent state may have been already created by old (earlier
    than 2.7) Mercurial and forgotten in repositories

    I actually know the user who hit forgotten inconsistent rebase
    state

  - switching back to old Mercurial temporarily is not so easy for all
    users

    in many cases, re-installation is needed

Patch #1 of this series was once posted as a part of the series to fix
problems of histedit.

  http://selenic.com/pipermail/mercurial-devel/2013-August/053095.html