[MERGE][0.15] Use a set during iter_entries_by_dir

[John Arbash Meinel]

This was a performance regression, especially on large trees. The big
difference is just changing the list to be a set, which has much better
characteristics for 'x in y' when the number of entries is large.

John
=:->
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: file_id_set.patch
Url: https://lists.ubuntu.com/archives/bazaar/attachments/20070302/622122c4/attachment.diff