logpr = new(logpr_gauss, om, terms)
This is a standard model of coefficients which has them as drawn
independently
from
$$ \theta_i \sim N(0, \rho c_i)$$
where \(c_i\) is the variance supplied by om for the \(i\)th term.
The parameter vector is of length 1 where
\(\rho=\) exp(para[0]).
an outermod instance to be referred to
a matrix of terms, must have as many columns as dims in
om
no returns, this is a class which contains methods
base class: lpdf