Fix LM surrogate operator documentation

src/plans/nonlinear_least_squares/nls_general_plan.jl

@@ -203,15 +203,15 @@ For the derivation of the Riemannian case, see [BaranBergmann:2026](@cite).
 
 In order to build a surrogate also for the robustified Levenberg-Marquardt, introduce
 ``α = 1 - $(_tex(:sqrt, "1 + 2 $(_tex(:frac, "ρ''(p)", "ρ'(p)"))$(_tex(:norm, "F(p)"; index = "2"))^2"))``
-and set ``y = $(_tex(:frac, _tex(:sqrt, "ρ'(p)"), "1-α"))F(p)`` and ``$(_tex(:Cal, "L"))(X) = CJ_F^*(p)[F(p)]``
+and set ``y = $(_tex(:frac, _tex(:sqrt, "ρ'(p)"), "1-α"))F(p)`` and ``$(_tex(:Cal, "L"))(X) = CJ_F(p)[X]``
 with
 
 ```math
 C = $(_tex(:sqrt, "ρ'(p)"))(I-αP), $(_tex(:qquad)) P = $(_tex(:frac, "F(p)F(p)^" * _tex(:rm, "T"), _tex(:norm, "F(p)"; index = "2") * "^2")),
 ```
 
-where ``F(p) ∈ ℝ^n`` is the vector of residuals at point ``p ∈ M`` and ``J_F^*(p): ℝ^n → $(_math(:TangentSpace))``
-is the adjoint Jacobian.
+where ``F(p) ∈ ℝ^n`` is the vector of residuals at point ``p ∈ M`` and ``J_F(p): $(_math(:TangentSpace)) → ℝ^n``
+is the Jacobian.
 These two can be accessed with [`get_vector_field`](@ref) for ``y`` and [`get_linear_operator`](@ref) for ``$(_tex(:Cal, "L"))``,
 respectively.
 For technical details on the scaling using ``α``, especially how the `threshold` and `mode`