[MERGE] knit text adding reductions

[Robert Collins]

This does:

CommitBuilder now knows if the added revision id is random or not, and
uses that down to the knit layer to reduce queries into the existing
index - during a commit with a random id these are all miss-cases, which
is the worst case for an index.

Knits no longer assert that content being inserted is correctly split
and bytestrings. The overhead of doing that was nearly 1/3 the time it
takes to gzip the content - a very high overhead, and one that is
unlikely to find bugs in bzr's core code. I've added docstring changes
to make it clear that this is not enforced by the API.

With these changes and non-annotated texts I'm down to 1m47 to do a
commit of mozilla on my laptop. tar czf is 38 seconds for comparison.

-Rob
-- 
GPG key available at: <http://www.robertcollins.net/keys.txt>.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: bzr-mail-UM3R5V.patch
Type: text/x-patch
Size: 20805 bytes
Desc: not available
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20070910/c9dc3fac/attachment-0001.bin 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20070910/c9dc3fac/attachment-0001.pgp