README and API improvements

[hackergrrl]

Adds a fancy-pants README, and also makes path and dagService non-optional parameters.

Addresses #9

[hackergrrl]

This test was already not passing as of master.

[daviddias]

It was failing because git doesn't check in empty dirs. @nginnever has pushes a fix to his branch https://github.com/ipfs/js-ipfs-data-importing/pull/6/files which essentially creates the empty folders if they are not there..

[daviddias]

@noffle could you redo this PR with just the README improvs. @nginnever PR will be merged soon which adds the expected behaviour on the test. Sorry that you had to go through that trouble.

[daviddias]

@noffle #6 was merged, you should see the nested dir test passing on master now. Could you redo your PR with the readme changes? Thank you :)

[hackergrrl]

Changes made. I'm also multiplexing on the target param instead of having options.{buffer|path|stream} and the complexity that introduces, since we only ever want one of these.

[hackergrrl]

Blocked on #10 for now.

[daviddias]

greeeen #12 :)

[hackergrrl]

The js-ipfs-merkle-dag changes don't make a difference for me -- Chrome and Firefox both still crash /wo logs.

[hackergrrl]

Tests are passing!

The home-made chunker used was allocating ~O(100n) bytes, causing the 1.2mb file to crash Chrome and Firefox. I switched us to block-stream2, which is O(n) and supports zero-padding too.

[daviddias]

woooo :)

[nginnever]

Removing chunker test file removed the node environment testing of the fixed size chunker, the test now only runs against the browser.

Is this desired?

The test wasn't set up to completely reduce redundancy and the same tests were in two different places and required separately instead of two requires to the same test file like in merkle-dag

[daviddias]

@nginnever missed that. We don't need to test the chunker anymore cause we are using block-stream2 which does it well for us. @noffle mind cleaning that part as well? Thank you both :)