Fix book build failure in machine_learning chapter caused by duplicate predictor names

[Copilot]

The build-book (devel) Actions job failed while rendering inst/pages/machine_learning.qmd in the Random Forest classification section. mikropml::run_ml() received a dataset with duplicate feature names, which caused dplyr::pull() to abort.

• Problem surfaced in rendered ML workflow

  • Failure occurred during supervised model training (run_ml) after preprocessing, due to non-unique feature/predictor names in the loaded TSE object.

• Targeted fix in chapter setup

  • Added a single normalization step immediately after loading the QinJ_2012 dataset to enforce unique taxonomic row names before downstream preprocessing/modeling.
  • This prevents duplicate columns when model inputs are materialized.

• Change

  # Ensure taxonomic feature names are unique for model matrix creation
  rownames(tse) <- make.unique(rownames(tse))