Clusto-apiserver

New release only to deal with dumb pypi upload error hurr. I basically futzed the upload.

- The `/by-attr` endpoint, which translates to the main clusto query method
`get_by_attr()`
- Pagination implementation for results. By default it is turned off, basically
only useful for those cases where your data is still too much. One important
caveat here is that this only affects the **presentation** of the data back
to the client, not the **querying** of the data from the database. Pagination
is not native to clusto, yet
- Minify implementation for results. All results by default are _pretty
printed_ but if you pass a request header they will be printed as a single
string

- Adding `requirements.txt`, `dev-requirements.txt` and
`test-requirements.txt` to the distribution tarball for
source downloads

- Split install requirements into 3 files, which can be read either by `pip`
or by the `setup.py` script.
- `requirements.txt` is the usual file you install from
- `test-requirements.txt` is a file intended for use when you are testing
this repo.
- `dev-requirements.txt` is the file you should install from if you are
hacking in this repo.
- Added a custom `Clusto-Mode` header to change how the JSON data is
presented back:
- `compact` mode: is the default for API calls that return a list of items
back to the requestor.
- `expanded` mode: is the default for API calls that return a single item
back to the requestor
- Added a `by-names` endpoint, it receives a list of parameters and will
return back a list of items found