docs: use fast lambdify in usage notebook step 3

Reverts a change introduced in #359

docs/usage/step3.ipynb

@@ -53,7 +53,7 @@
    "source": [
     "As explained in the {doc}`previous step <step2>`, a {class}`.ParametrizedFunction` can compute a list of intensities (real numbers) for an input {obj}`.DataSample`. At this stage, we want to optimize the parameters of this {class}`.ParametrizedFunction`, so that it matches the distribution of our data sample. This is what we call 'fitting'.\n",
     "\n",
-    "First, we load the relevant data from the previous steps."
+    "First, we load the relevant data from the previous steps. Notice that we use {func}`.create_parametrized_function` with the argument `max_complexity`, which speeds up lambdification (see {doc}`/usage/faster-lambdify`)."
    ]
   },
   {
@@ -89,6 +89,7 @@
     "    expression=model.expression.doit(),\n",
     "    parameters=model.parameter_defaults,\n",
     "    backend=\"jax\",\n",
+    "    max_complexity=100,\n",
     ")"
    ]
   },