Handle brms 0 + Intercept formulas in get_modelmatrix()

[Copilot]

report() could fail on brms models fitted with formulas like 0 + Intercept + wt because insight::get_modelmatrix() reconstructed the RHS from variable names and treated Intercept as if it were a data column. This updates the brms path to preserve the intercept correctly when rebuilding the model matrix.

• Model-matrix reconstruction

  • Update get_modelmatrix.brmsfit() to:
    • detect intercept presence separately
    • exclude symbolic Intercept from predictor extraction
    • rebuild the formula with the correct intercept setting
  • keep intercept-only cases working without requiring an Intercept column in the data

• Regression coverage

  • Add a focused brms test for formulas of the form 0 + Intercept + wt
  • Assert the resulting model matrix has the expected columns:
    • "(Intercept)"
    • "wt"

• Package metadata

  • Bump the development version
  • Add a NEWS entry for the user-visible bug fix

Example of the affected case:

m <- brms::brm(mpg ~ 0 + Intercept + wt, data = mtcars, refresh = 0)

insight::get_modelmatrix(m)
# now returns columns: "(Intercept)", "wt"