pub fn make_classify_documents_task() -> TypedTask<CognifyInput, ClassifiedDocuments>Expand description
Build a TypedTask that classifies Data items into Documents.
The returned task does not carry a name; the pipeline builder
build_cognify_pipeline wraps it with CLASSIFY_DOCUMENTS_TASK_NAME.
In-body provenance stamping: stamps every emitted Document with
source_pipeline = "cognify" and source_task = "classify_documents".
Necessary because ClassifiedDocuments is a non-HasDataPoint wrapper
not walked by the executor’s stamp_tree_dyn (LIB-06-03 fixup).