[MERGE] Implement Tree.plan_merge, deprecate RevisionTree.get_weave

[Robert Collins]

Robert Collins has voted +1 (conditional).
Status is now: Conditionally approved
Comment:

+            # Disable pending merges, to avoid affecting merge behavior
+            tree.set_parent_ids(parents[:1])

Could you explain more why this is needed? It seems prone to losing a 
users data e.g. during a power-off or kill -9 situation as well as 
requiring more work at the tree layer to discard and recreate the cached 
parent data.

with our current API

+    def _get_ancestors(self, default_revision):
+        ancestors = set([default_revision])
+        for parent_id in self.get_parent_ids():
+            ancestors.update(self.branch.repository.get_ancestry(
+                             parent_id, topo_sorted=False))
+        return ancestors

will be faster as
+        ancestors = set([default_revision])
+ 
ancestors.update(self.branch.repository.get_revision_graph(self.get_parent_ids()))
+        return ancestors

because that will do just one index traversal.

Its interesting that we grab full ancestry here. This doesn't affect the 
merge, but I wonder about giving a callback here, or even using the file 
merge graph which will be much smaller to reduce the amount of index 
looked at.

With the previously discussed plan_merge -> plan_file_merge I'm happy to 
see this come in.

For details, see: 
http://bundlebuggy.aaronbentley.com/request/%3C469D2641.7020504%40utoronto.ca%3E