There are a ton related to performance, which you can probably skip for now. Those should be reviewed with a fine-toothed comb, since most performance fixes make the code harder to read, unfortunately.
This one is kind of a big deal for pouchdb-server. Basically I wanted to set it up so that when we test against pouchdb-server, we don't cram all the files in the root directory (annoying during development), and in the process I discovered a bug in how the db was reporting its name in db.info. This fix makes sure that the name is the same everywhere (on('destory'), on('create'), db.info()) and is something sensible, regardless of the prefix.
More pouchdb-server stuff. I think it makes sense to go ahead and merge this, even though it's failing. Gives us an excuse to make more improvements to MemDOWN.
(edit depends on the previous PR, also on a commit that has already been merged, so you can just cherry-pick the final commit)
Simple fix for Cloudant, basically does the thing you suggested somewhere wher ewe ensure that the response to bulkDocs/put/post is always {ok: true} even if the backend is Cloudant. Tested manually and it fixes the test.basics.js against Cloudant.
You can probably review this pretty quickly. It's basically a follow-up to the _local docs changes to fix the update_seq. That bug that was reported about changes() could also apply in the somewhat rare situation of new_edits; this adds a test to repro his issue and also fixes it. Does some tricky stuff with managing transactions in idb/websql, but it definitely works.