[Python] Fix TF1 Pythonization test for NumPy 2.4.0

The new NumPy 2.4.0 is more strict when implicitly converting 1-element
arrays to scalars. It doesn't do that anymore, causing the TF1
Pythonization tests to fail (see log below).

This actually pointed to a real mistake in setting up the test, where a
2D array was used to define the TFormula parameters while it should be a
1D array.

```txt
 962/3718 Test   #93: pyunittests-bindings-pyroot-pythonizations-pyroot-pyz-tf-pycallables ..............................***Failed    3.05 sec
test_callable (tf_pycallables.TF1.test_callable)
Test function provided as callable ... ok
test_evalpar (tf_pycallables.TF1.test_evalpar)
Test the 2D Numpy array pythonizations for TF1::EvalPar ... ERROR
test_evalpar_dynamic (tf_pycallables.TF1.test_evalpar_dynamic)
Test the 2D NumPy pythonizations with dynamic TF1 data dimensions ... ok
test_fitgauss (tf_pycallables.TF1.test_fitgauss)
Test fitting a histogram to a Python function ... ok
test_identity (tf_pycallables.TF1.test_identity)
Test simple function without parameters ... ok
test_params (tf_pycallables.TF1.test_params)
Test function with parameters ... ok
test_params (tf_pycallables.TF2.test_params)
Test function with parameters ... ok
test_params (tf_pycallables.TF3.test_params)
Test function with parameters ... ok

======================================================================
ERROR: test_evalpar (tf_pycallables.TF1.test_evalpar)
Test the 2D Numpy array pythonizations for TF1::EvalPar
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/github/home/ROOT-CI/src/bindings/pyroot/pythonizations/test/tf_pycallables.py", line 129, in test_evalpar
    expected_value = pyf_tf1_coulomb(x[i, ::2], params)
  File "/github/home/ROOT-CI/src/bindings/pyroot/pythonizations/test/tf_pycallables.py", line 33, in pyf_tf1_coulomb
    return p[1] * x[0] * x[1] / (p[0]**2) * math.exp(-p[2] / p[0])
                                            ~~~~~~~~^^^^^^^^^^^^^^
TypeError: only 0-dimensional arrays can be converted to Python scalars

----------------------------------------------------------------------
Ran 8 tests in 1.469s

FAILED (errors=1)
CMake Error at /github/home/ROOT-CI/src/cmake/modules/RootTestDriver.cmake:232 (message):
  error code: 1

```

(cherry picked from commit d5135f7)

Author: guitargeek

bindings/pyroot/pythonizations/test/tf_pycallables.py

@@ -116,11 +116,13 @@ def test_evalpar(self):
             [3.0, 10, 4.0]
         ])
 
-        params = np.array([
-            [1.0],       # Distance between charges r
-            [8.99e9],    # Coulomb constant k (in N·m²/C²)
-            [0.1]        # Additional factor for modulation
-        ])
+        params = np.array(
+            [
+                1.0,  # Distance between charges r
+                8.99e9,  # Coulomb constant k (in N·m²/C²)
+                0.1,  # Additional factor for modulation
+            ]
+        )
 
         # Slice to avoid the dummy column of 10's
         res = rtf1_coulomb.EvalPar(x[:, ::2], params)