Use "spawn" start method for multiprocessing

This avoids issues caused by mixing forking and threading in the same
process. The default for multiprocessing will be changed to "spawn"
in python 3.14, but this prevents warnings from being emitted until
then.

src/CSET/operators/ageofair.py

@@ -404,6 +404,10 @@ def compute_ageofair(
     logging.info("STARTING AOA DIAG...")
     start = datetime.datetime.now()
 
+    # Use "spawn" method to avoid warnings before the default is changed in
+    # python 3.14. See the (not very good) warning here:
+    # https://docs.python.org/3/library/multiprocessing.html#contexts-and-start-methods
+    mp_context = multiprocessing.get_context("spawn")
     # Main call for calculating age of air diagnostic
     if ensemble_mode:
         for e in range(0, len(XWIND.coord("realization").points)):
@@ -425,7 +429,7 @@ def compute_ageofair(
                 tmpdir.name,
             )
             if multicore:
-                with multiprocessing.Pool(num_usable_cores) as pool:
+                with mp_context.Pool(num_usable_cores) as pool:
                     pool.map(func, range(0, XWIND.shape[4]))
             else:
                 # Convert to list to ensure everything is processed.
@@ -452,7 +456,7 @@ def compute_ageofair(
             tmpdir.name,
         )
         if multicore:
-            with multiprocessing.Pool(num_usable_cores) as pool:
+            with mp_context.Pool(num_usable_cores) as pool:
                 pool.map(func, range(0, XWIND.shape[3]))
         else:
             # Convert to list to ensure everything is processed.