[MERGE] Actually Deprecate Unicode file + revision ids

[John Arbash Meinel]

The attached patch changes 'safe_revision_id' and 'safe_file_id' to emit
a warning if given a Unicode string. And then I track down a few places
in the code that were still generating Unicode ids. IIRC there was only
one that I had missed in the past (merge_modified). The other cases are
 just cleanup, or updating the test suite.

Tests that were testing a Unicode id now generally have 1 test which
checks the deprecation is raised, and then another test that checks for
utf8 ids.

I also did a couple small updates which should lessen the cost of
calling these functions since bzr.dev is currently slower than 0.14 in
some cases because of these checks.

I've audited our code base for them, so the only places that we have to
be careful are for the converters. And baz-import seems to pass all
tests, and I know cvsps-import explicitly uses ascii file ids and
revision ids. The only one I'm not sure on is bzr-svn.

John
=:->
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: unicode_id_warnings.patch
Url: https://lists.ubuntu.com/archives/bazaar/attachments/20070302/9219e6d3/attachment-0001.diff