Register a plugin engine for fmri_lm — register

Usage

register_engine(name, fit, preflight = NULL, capabilities = list())

name: Character scalar identifier advertised to users (e.g. "friman").
fit: Function invoked as fit(model, dataset, args, cfg) and expected to return an fmri_lm object.
preflight: Optional function invoked before fitting; receives the same arguments as fit and can signal errors early.
capabilities: Optional named list describing engine support for global fmri_lm() options. Recognized fields currently include robust, preprocessing, ar_voxelwise, ar_by_cluster, plus contextual rules such as requires_event_regressors, requires_parcels_for_by_cluster, and forbid_by_cluster_dataset_classes.

Invisibly, TRUE.