`stan_mvmer` fails when two random intercepts are specified

Summary:

stan_mvmer fails when two grouping factors are specified both with just a random intercept.

Reproducible Steps:

The following model fails:

dat$herd2[dat$herd %in% 1:5] = 1
dat$herd2[dat$herd %in% 6:10] = 2
dat$herd2[dat$herd %in% 11:15] = 3

mod <- 
  stan_mvmer(incidence ~ size + period + (1|herd) + (1|herd2),
             data = dat, 
             chains = 1,
             seed = 12345,
             refresh = 0,
             prior_covariance = lkj())

with an error about dimensions in the priors

Error in new_CppObject_xp(fields$.module, fields$.pointer, ...) : 
  Exception: mismatch in number dimensions declared and found in context; processing stage=data initialization; variable name=b2_prior_regularization; dims declared=(); dims found=(0)  (in '/data/hyperparameters_mvmer.stan' at line 42; included from 'model_mvmer' at line 29)

So does a model with the decov() prior (interestingly, with a different error, but still about prior dimensions):

mod <- 
  stan_mvmer(incidence ~ size + period + (1|herd) + (1|herd2),
             data = dat, 
             chains = 1,
             seed = 12345,
             refresh = 0,
             prior_covariance = decov())

with error

Error in new_CppObject_xp(fields$.module, fields$.pointer, ...) : 
  Exception: mismatch in dimension declared and found in context; processing stage=data initialization; variable name=b_prior_shape; position=0; dims declared=(2); dims found=(1)  (in '/data/hyperparameters_mvmer.stan' at line 31; included from 'model_mvmer' at line 29)

The automated tests use additional random effects (two random intercepts + one random slope) which seems to work ok, hence why the tests didn't fail. It is only when it is two random intercepts and no additional random effects that it fails. For example, the following seems to work:

mod <- 
  stan_mvmer(incidence ~ size + period + (1|herd) + (period|herd2),
             data = dat, 
             chains = 1,
             seed = 12345,
             refresh = 0)

RStanARM Version:

2.19.3

R Version:

3.6.2

Operating System:

Windows 10