Per process shared memory

Currently import data is stored in shared memory with the keys being the hash value of the respective file paths. This leads to the problem, that when running more than one Discretizer (or executing more than one Discretizer docker container) at the same time, whichever Discretizer process finishes first closes the shared memory causing the remaining Discretizer to lose access to it.

This MR changes the hashed key to include the processed quadkeys, so each shared memory instance is unique to the file path and the processed quadkeys.

\approve @ds @laurens

discretizer/datamanager.py

@@ -169,7 +169,9 @@ class DataManager(TilesDatabase, Quadtile):
         _quadkeys_str = "-".join(_quadkeys)
 
         # Return a hash value unique to the Discretizer process
-        return hashlib.md5(os.path.realpath(f"{filepath}/{_quadkeys_str}").encode("utf-8")).hexdigest()
+        return hashlib.md5(
+            os.path.realpath(f"{filepath}/{_quadkeys_str}").encode("utf-8")
+        ).hexdigest()
 
     def features(self, filepath, **kwargs):
         """