[BUG] Fix measure from sample shot vector condition (#1981)

**Context:**
With the new `shots` param added to the test
`test_measurement_from_samples_single_measurement_analytic`, it fails
immediately. The root cause has been discussed in a [PennyLane
discussion](PennyLaneAI/pennylane#7317 (comment)).

**Description of the Change:**
Use `shots.has_partitioned_shots` instead of `len(shots.shot_vector) >
1`

**Benefits:**

**Possible Drawbacks:**

**Related GitHub Issues:**
[sc-91172]

Author: JerryChen97

doc/releases/changelog-dev.md

@@ -56,6 +56,9 @@
 
 <h3>Bug fixes 🐛</h3>
 
+* Fix wrong handling of partitioned shots in the decomposition pass of `measurements_from_samples`.
+  [(#1981)](https://github.com/PennyLaneAI/catalyst/pull/1981)
+
 * Fix errors in AutoGraph transformed functions when `qml.prod` is used together with other operator
   transforms (e.g. `qml.adjoint`).
   [(#1910)](https://github.com/PennyLaneAI/catalyst/pull/1910)

frontend/catalyst/device/decomposition.py

@@ -348,7 +348,7 @@ def postprocessing_samples(results):
         results_processed = []
         for m in tape.measurements:
             if isinstance(m, (ExpectationMP, VarianceMP, ProbabilityMP, SampleMP)):
-                if len(tape.shots.shot_vector) > 1:
+                if tape.shots.has_partitioned_shots:
                     res = tuple(m.process_samples(s, measured_wires) for s in samples)
                 else:
                     res = m.process_samples(samples, measured_wires)

frontend/test/pytest/test_measurement_transforms.py

@@ -478,7 +478,7 @@ def circuit(theta: float):
             ),
         ],
     )
-    @pytest.mark.parametrize("shots", [3000, (3000, 4000), (3000, 3500, 4000)])
+    @pytest.mark.parametrize("shots", [3000, (3000, 3000), (3000, 4000), (3000, 3500, 4000)])
     def test_measurement_from_samples_single_measurement_analytic(
         self,
         input_measurement,