Highway RNN builder

A highway LSTM ([Srivastava et al., 2015](https://papers.nips.cc/paper/5850-training-very-deep-networks.pdf); [Zhang et al., 2016](https://groups.csail.mit.edu/sls/publications/2016/YuZhang3_ICASSP-16.pdf); [He et al., 2017](http://aclweb.org/anthology/P/P17/P17-1044.pdf)) is similar to a deep BiLSTM, but each layer's output is linearly combined with its input before being passed on to the next layer, where the combination is determined by a gate depending on the previous time step and the previous layer's output. This alleviates vanishing gradients along the layers.
A highway RNN builder shouldn't be much harder than the existing BiRNN builder to implement.

Start by reading the cited highway-network papers and locating the existing BiRNN builder in the DyNet codebase. Compare its interface and behavior with the requested highway LSTM design; done means a highway RNN builder is implemented with the described gated layer connections and validated against the existing builder's conventions.

Written by the indexing model from the issue text.