Add transfer entropy

[galengorski]

This PR adds to the it_functions.py library. The central function is calcTE which calculates the transfer entropy between source and sink with respect to a time lag chosen using the parameter "shift". Transfer entropy is calculated using the Knuth (2005) formulation (see equation 14). Other functions added are:

• calc3dpdfs : calculates the 3-d joint pdf between variables
• lag_data : creates an array of lagged data based on the input parameter "shift"
• calcTE_shuffled: shuffles the input data to destroy temporal dependence and then calculates TE, for significance testing
• calcTE_crit: uses output from calcTE_shuffled and input significance thresholds to estimate a significance threshold for TE, above which TE values are considered "significant"

[galengorski]

Thanks for the review @jds485 , sorry there were other functions in the PR that weren't related to it_analysis.py but thanks for focusing on that