The Effects of Random and Discrete Sampling when Estimating Continuous-Time Diffusions