replace vs. delete causing TC

Hi tree-conflicts folks,

if you might, take a look at merge_tests.py 135
(merge_replace_causes_tree_conflict2) in r39132.

This test is the same as 132, but it doesn't locally modify, it locally
deletes. Technically they all should act the same as 132, but see for yourself.

The comments starting with '###' show all the current failures.
(see merge_replace_setup()'s top comment showing the commands to see which
path is which replace kind)

1. A/mu -- file replaced with file
Fails to register as a TC during merge notification. Says both 'Skipped' and
'A'. (status notes the TC) -- Should be relatively easy to fix, I guess.

2. A/D/G/pi -- file replaced with dir
Plain refuses to merge the replacement, says 'delete needs to be committed
first'. -- No idea how to fix, but it works well in all other cases!

3. In the end, the status should be like 'R C ', but

  a) A/mu and A/B/E have a '+' status as well, although it isn't a
     replace-with-history. (Still unsure about the full meaning of '+')

  b) A/D/H's status is 'D ', not 'R '.

( c) A/D/G/pi is missing because of point 2. )

~Neels

------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=2390889