This interface avoids leaking transport-specific information, while
returning an indication of how the request fared so the user can decide
on how to proceed with processing.

Instead of mixing the Batch model for batches sent *to* with batches
received *from* SETRA, we switch to using an InputBatch and an
OutputBatch. The InputBatch contains the fields that can be used when
sending a Batch (with Vouchers and Transactions), and the OutputBatch
contains all the information one receives from SETRA when fetching a
single batch from the batch/ endpoint, or multiple using the search
functionality.

The get_batch method is reduced to only fetch a single batch, while the
new search_batches method takes the role of searching based on creation
date and interface. In addition, the return_list_of_obj parameter is
removed in favor of the return_objects class attribute.

This commit expects that SETRA includes the id field in Batches.

- Both get_batch_errors and get_batch_complete now return objects
  unless the client parameter return_objects is set to False
- Tests for models, endpoints and client methods are included.

Also updated and fixed tests, fixtures and readme

And added two new tests

Introduces SetraClient. The client is made
for use against the SETRA(SEntralt TRAnsaksjonslager) API.

There are models for Batch, Voucher, Transaction and Multi, where Multi is
a combination of the preceding, The Multi model validates that the batchid
of the voucher matches the batch included, and that the voucherid of the
transactions matches one of the included vouchers.

Tests have also been included for the client itself, the endpoints, and
the models.

Configuration for tox, Jenkins and gitlab-ci are included for running
tests and building the client.