sample_intercept: Sample intercept effects
In zhifeiyan/WeibullHM: A Gibbs Sampler of a Bayesian Weibull Hidden Markov Model

Produce a Gibbs sample of intercept effects of all states

1	sample_intercept(y, alpha, var_logscale, m_mu, sigma_mu_inv, nsubj)

`y`	a matrix of current update of Weibull log scale parameters
`alpha`	a matrix of current update of subject random effects
`var_logscale`	a vector of current update of variances of Weibull log scale parameters
`m_mu`	a vector of means of multivariate normal prior of intercept effects
`sigma_mu_inv`	inverse covariance matrix of multivariate normal prior of intercept effects
`nsubj`	total number of subjects