Fix failing test_specs test cases for legacy opmath (#5871)

**Context:**
Latest failures with legacy opmath:
https://github.com/PennyLaneAI/pennylane/actions/runs/9557831967/job/26346373805

**Description of the Change:**
- Replaces `qml.X(0) + qml.Y(1)` with `qml.sum(qml.X(0), qml.Y(1))`
because the `Sum` here does not have `data` but `Hamiltonian` does (it
treats the two coefficients of `1` and `1` as data), which caused
`test_int_specs_level` to fail at verifying the number of parameters.
- Adds a `__len__` to the legacy `Hamiltonian` such that `assert
len(specs_list) == len(H)` in `test_splitting_transforms` passes
correctly.

[sc-66284]

pennylane/ops/qubit/hamiltonian.py

@@ -318,6 +318,10 @@ def __init__(
         super().__init__(*coeffs_flat, wires=self._wires, id=id)
         self._pauli_rep = "unset"
 
+    def __len__(self):
+        """The number of terms in the Hamiltonian."""
+        return len(self.ops)
+
     @property
     def pauli_rep(self):
         if self._pauli_rep != "unset":

tests/resource/test_specs.py

@@ -37,7 +37,7 @@ def circuit(x):
             qml.SWAP((0, 1))
             qml.X(0)
             qml.X(0)
-            return qml.expval(qml.X(0) + qml.Y(1))
+            return qml.expval(qml.sum(qml.X(0), qml.Y(1)))
 
         return circuit