subject offset initvals do not correspond to sigma in initial_point

[frankmj]

Describe the bug

The initvals in initial_point for participant offsets are all very close to 0 even if the sigma for the corresponding distribution is high. Shouldn't they be initialized according to the distribution implied by the inits of mu and sigma?

HSSM version
What is the version of HSSM you are using?

0.2.3 but also development

To Reproduce

in model specification:

"formula": "v ~ 1 + (1|participant_id)"

"prior": {
"Intercept": {"name": "Normal", "mu": 1, "sigma": 2, "initval": 1},
"1|participant_id": {"name": "Normal",
"mu": 0, #
"sigma": {"name": "HalfNormal",
"sigma": 5,
"initval": 3}
{

then look at initial_point() one can see the sigma is set with initval 3 as desired but the participant offsets are very small and are insensitive to changes in sigma

Screenshots