[Tests] Fix Autograph warnings in division test (#1329)

**Context:** Currently, there are two `UserWarning`s issued in the test
`TestJaxIndexOperatorUpdate::test_single_index_div_update_all_items`:

Worflow `f`:

```
frontend/test/pytest/test_autograph.py::TestJaxIndexOperatorUpdate::test_single_index_div_update_all_items
  .../catalyst/frontend/catalyst/autograph/ag_primitives.py:350: UserWarning: Tracing of an AutoGraph converted for loop failed with an exception:
    AutoGraphError:    The variable 'x' was initialized with the wrong type, or you may be trying to change its type from one iteration to the next. Expected: ShapedArray(float64[3]), Got: ShapedArray(int64[3])
  
  The error ocurred within the body of the following for loop statement:
    File ".../catalyst/frontend/test/pytest/test_autograph.py", line 2165, in f
      for i in range(first_dim):
  
  If you intended for the conversion to happen, make sure that the (now dynamic) loop variable is not used in tracing-incompatible ways, for instance by indexing a Python list with it. In that case, the list should be wrapped into an array.
  To understand different types of JAX tracing errors, please refer to the guide at: https://jax.readthedocs.io/en/latest/errors.html
  
  If you did not intend for the conversion to happen, you may safely ignore this warning.
    warnings.warn(
```

Workflow `g`:

```
frontend/test/pytest/test_autograph.py::TestJaxIndexOperatorUpdate::test_single_index_div_update_all_items
  .../catalyst/frontend/catalyst/autograph/ag_primitives.py:350: UserWarning: Tracing of an AutoGraph converted for loop failed with an exception:
    AutoGraphError:    The variable 'result' was initialized with the wrong type, or you may be trying to change its type from one iteration to the next. Expected: ShapedArray(float64[3]), Got: ShapedArray(int64[3])
  
  The error ocurred within the body of the following for loop statement:
    File ".../catalyst/frontend/test/pytest/test_autograph.py", line 2176, in g
      for i in range(first_dim):
  
  If you intended for the conversion to happen, make sure that the (now dynamic) loop variable is not used in tracing-incompatible ways, for instance by indexing a Python list with it. In that case, the list should be wrapped into an array.
  To understand different types of JAX tracing errors, please refer to the guide at: https://jax.readthedocs.io/en/latest/errors.html
  
  If you did not intend for the conversion to happen, you may safely ignore this warning.
    warnings.warn(
```

Both warnings are symptoms of the same issue: the input is an array of
integers, `np.array(5, 3, 4)`, but the test is performing division
(directly on the input in worflow `f` and on a copy of the input in
worflow `g`), which requires changing the array's type to floats, hence
the warning.

**Description of the Change:** Changes the input to an array of floats:
`np.array([5.0, 3.0, 4.0])`. This fixes the warnings since there is now
no array type conversion during the division operation.

**Benefits:** Fixes warnings in test suite; cleaner test output.

**Possible Drawbacks:** None.

frontend/test/pytest/test_autograph.py

@@ -2182,7 +2182,7 @@ def g(x):
             expected = g(x)
             return result, expected
 
-        result, expected = workflow(np.array([5, 3, 4]))
+        result, expected = workflow(np.array([5.0, 3.0, 4.0]))
         assert jnp.allclose(result, jnp.array([2.5, 1.5, 2]))
         assert jnp.allclose(result, expected)