sigma_E was coming out as a vector (because of the time-varying option) and the model component subsetting was then messed up with tidy() and therefore print() such that the positive component sigma_E was reported as the binomial component sigma_E.
This only affected printing/tidy() of delta models with spatiotemporal fields.
We might get a slight speed boost by only sdreporting one log_sigma_E if sigma_E isn't time varying.
Fix: https://github.com/pbs-assess/sdmTMB/commit/5334b1dea618fecd1ce88f73d4e3053c6088b964
sigma_E was coming out as a vector (because of the time-varying option) and the model component subsetting was then messed up with tidy() and therefore print() such that the positive component sigma_E was reported as the binomial component sigma_E.
This only affected printing/tidy() of delta models with spatiotemporal fields.
We might get a slight speed boost by only sdreporting one log_sigma_E if sigma_E isn't time varying.