least_squares_problem
=====================

.. py:function:: t3toolbox.backend.optimizers.least_squares_problem(geom, kind, sample, data, regularizer = None)

   .. code-block:: python

      def least_squares_problem(
              geom:        GeometryOps,   # COREWISE_OPS / MANIFOLD_OPS
              kind:        typ.Any,       # backend fitting.{APPLY,ENTRIES,PROBE} or a derivative kind
              sample:      typ.Any,       # ww / index / (ww,pp) / (index,pp)
              data:        typ.Any,       # observed values
              regularizer: typ.Any = None,  # optional backend.regularization.Regularizer (e.g. IdentityRegularizer(λ))
      ) -> Problem:


   Assemble a least-squares ``Problem`` from a geometry, a sampling kind, the sample, and the observed
   data, plus an optional ``regularizer`` (adds ``ρ(x)`` to the objective, folded into the local GN model).
   (The frontend adapter calls this with the same backend objects a raw-data user would.)