Stop listing after `stream` atoms

Nobody has claimed this yet.

1MiB MP4 Sample: http://www.sample-videos.com/video/mp4/240/big_buck_bunny_240p_1mb.mp4

MP4 structure:

> ftyp (32 bytes)
> free (8 bytes)
> mdat (1046699 bytes)
> moov (6904 bytes)

Code snippet:

``` javascript
var mp4 = require('mp4-stream')
var fs = require('fs')

var decode = mp4.decode()

fs.createReadStream('video.mp4')
.pipe(decode)
.on('box', function (box, next) {
console.log('found box (' + box.type + ') at ' + box.offset + ' (' + box.length + ')')
if (box.container) {
console.log('box has child boxes (will be emitted next)')
}
if (box.stream) {
console.log('box has stream data (consume stream to continue)')
box.stream.resume()
}
next() // call this when you're done
})
```

Result:

> found box (ftyp) at 0 (32)
> found box (free) at 32 (8)
> box has stream data (consume stream to continue)

Stop listing after `free` atom.

Reproduce the issue with the supplied 1MiB MP4 sample and the shown mp4.decode() pipeline. Start at the decode entry point and trace the `box` callback after `box.stream.resume()` and `next()`. Done means the decoder continues listing boxes after the `mdat` stream, including the `moov` atom described in the issue.

Written by the indexing model from the issue text.