Adding Gaussian Process Covariance Functions


Sorry to revive this, but @avehtari is the jitter parameter in the covariance function definition a necessity? The # of test cases blows up with this specification, and an add_diag function in the library per @Bob_Carpenter’s suggestion will probably have more uses elsewhere, and it will save lots of time in implementation… apologies I didn’t get to this until today



I’m fine with this.