encoderModel_gaussian_kernel: Guassian kernel
In scAlign: An alignment and integration method for single cell genomics

Tensorflow implementation of tsne's gaussian_kernel, accepts tensors of float64 (for precision) and returns tensors of float32

1 2	encoderModel_gaussian_kernel(data, data_b = NULL, dim, perplexity = 30, method = "euclidean", betas.fixed = 0, diag = "zero")

`data`	cell x feature data matrix
`data_b`	second cell x feature data matrix, if none provided then data_b = data
`dim`	number of cells in of mini-batch
`perplexity`	neighborhood parameter for gaussian kernel
`method`	distance metric prior to computing probabilities
`diag`	indicator for self similarity