Add support for source factor models (awslabs#275)

Source factors are enabled by passing --source-factors file1 [file2 ...] (-sf), where file1, etc. are token-parallel to the source (-s).
This option can be passed both to sockeye.train or in the data preparation step, if data sharding is used.
An analogous parameter, --validation-source-factors, is used to pass factors for validation data.
The flag --source-factors-num-embed D1 [D2 ...] denotes the embedding dimensions.
These are concatenated with the source word dimension (--num-embed), which can continue to be tied to the target (--weight-tying --weight-tying-type=src_trg).

At test time, the input sentence and its factors can be passed by multiple parallel files (--input and --input-factors) or through stdin with token-level annotations, separated by |. Another way is to send a string-serialized JSON object to the CLI through stdin which needs to have a top-level key called 'text' and optionally a key 'factors' of type List[str].

* Cleanup of vocab functions

* Simplified vocab logic a bit. Removed pickle functionality since it has been deprecated for long

* Refactor so that first factor corresponds to the source surface form (e.g. configs by default set num_factors to at least 1)

* fixed a TODO. slightly reworded the changelog

* Reworked inference interface. Added a bunch of TranslatorInput factory functions (including json)

* Removed max_seq_len_{source,target} from ModelConfig

* Separate data statistics relevant for inference from data information relevant only for training.

* Bumped Major Version to 1.17.0

* Do not throw exceptions while translating (awslabs#294)

* Remove bias parameters in Transformer attention layers as they bring no benefit. (awslabs#296)

Added custom speedometer to exactly track samples/sec and words/sec d…

…uring training (awslabs#260)

Fix entry points in setup.py (awslabs#258)

Missing commata.

Update to mxnet==1.0 (awslabs#244)

* Fix inference dims for mxnet 1.0

* Improved mx.NDArray indexing: removed intermediate numpy arrays for scores and topk hyp/word indices

* Removed unused parameter

* Reformatting inference.py

* simplify _get_inference_input()

* Update dependencies to mxnet==1.0.0

* Expose nccl kvstore and gradient compression

* Update major version

Fixed the maximum input length calculation at inference. (awslabs#255)

* Fixed the maximum input length calculation at inference.

* doc string

Bugfix: --num-samples-per-shard must be int (awslabs#254)

* Bugfix: --num-samples-per-shard must be int

* bump version

Sharded data iterator. (awslabs#241)

* Sharded data iterator.

* Added remaining sockeye/*.py files to typechecked files (awslabs#242)

* Tests to see we get the right number of batches.

* Improved log message about vocabs a little bit

* Factored validation iter creation into separate function

* Covering prepare data in the system tests.

* Writing a data version.

Remove RNN parameter packing, FusedRNN support; refactored core model…

… components (awslabs#189)

* Removed RNN parameter packing and FusedRNN support

* Refactor embedding and output layers (awslabs#196)

* Removed RNN parameter packing and FusedRNN support

* Refactoring of sockeye model: source embed/target embed/output layers are now separate components in model

* Make training and inference work. Remove lexical biasing code.

Update to MXNet version 0.12.1. (awslabs#213)

Yet another fix for the data iterator. Added a test. (awslabs#188)

* Yet another fix for the data iterator. Added a test that would catch this kind of problem

* Bump minor version