split
Splits a long-format table into one numeric series per distinct id.
rows are the data rows only (no header). headers names each column in row-position order and is used to resolve idColumn and valueColumn to indices.
Ids preserve first-encountered order. Each group is materialized to a DoubleArray in the order rows were supplied.
Throws ImportException when: a required column name is not found in headers, a row is shorter than the resolved column indices, a value-column cell does not parse as a double, or a resulting group is empty (cannot occur unless callers pre-filter).
Splits parallel id / value columns into one numeric series per distinct id, preserving first-encountered id order. Used by the database reader, which has already-typed columns rather than string rows.
ids and values must have the same length. Throws ImportException when they do not, or when a resulting group is empty.