Inference-tools

This release contains significant improvements to the `GpRegressor` class, including:

- A new option to select between the squared-exponential and rational-quadratic covariance functions, or provide a user-defined custom covariance function.

- A new option to use leave-one-out cross-validation to select hyper-parameter values instead of the marginal-likelihood.

- Significant improvements to numerical efficiency leading to reduced computation times.

- Fixed various bugs that appeared when testing after updating dependencies to numpy 1.15.0, scipy 1.3.1 and matplotlib 3.1.1

- Added a new class `ParallelTempering` to the `inference.mcmc` module which runs the 'parallel tempering' MCMC algorithm.

- `ParallelTempering` uses Python's `multiprocessing` module to create separate, dedicated processes for each Markov-chain object involved in the parallel tempering algorithm, allowing the necessary computations for each object to take place in parallel if multiple CPU threads are available.

- Added a new method `GpRegressor.gradient`, which allows for the calculation of the mean and variance of the gradient of the regression estimate. This is useful for robust estimation of derivatives in noisy data.

- Added a new function trace_plot to inference.plotting

- fixed a serious bug in inference.mcmc.MarkovChain.matrix_plot

This release includes significant improvements to package documentation, which is now also hosted online via Read the Docs.