chore: accommodate numpy immutable scalar memory sharing (#2675)

With numpy 2.4.0, most scalars share memory numpy/numpy@5386540, relax some test comparisons to accommodate this

Author: wpbonelli

autotest/test_copy.py

@@ -39,7 +39,11 @@ def model_is_copy(m1, m2):
             return False
     for k, v in m1.__dict__.items():
         v2 = m2.__dict__[k]
-        if v2 is v and type(v) not in [bool, str, type(None), float, int]:
+        # Allow identity sharing for immutable types including NumPy scalars
+        is_immutable = type(v) in [bool, str, type(None), float, int] or isinstance(
+            v, np.generic
+        )
+        if v2 is v and not is_immutable:
             # some mf6 objects aren't copied with deepcopy
             if isinstance(v, MFSimulationData):
                 continue
@@ -78,7 +82,16 @@ def package_is_copy(pk1, pk2):
     """
     for k, v in pk1.__dict__.items():
         v2 = pk2.__dict__[k]
-        if v2 is v and type(v) not in [bool, str, type(None), float, int, tuple]:
+        # Allow identity sharing for immutable types including NumPy scalars
+        is_immutable = type(v) in [
+            bool,
+            str,
+            type(None),
+            float,
+            int,
+            tuple,
+        ] or isinstance(v, np.generic)
+        if v2 is v and not is_immutable:
             # Deep copy doesn't work for ModflowUtltas
             if not inspect.isclass(v):
                 return False