Temporary fix to ujson

Fixed ujson version to 1.35 for now. The newest version 2.0.1 that was released a few days ago seems to be broken as it fails to encode some things that were not a problem before.

Also slipped in a very small doc update for the CLI use cases.

docs/cli_use_cases.rst

@@ -21,7 +21,7 @@ the following CLI command inside the target deployment:
 
 .. code-block:: sh
 
-    nomad client -n <api_url> -u <username> -w <password> mirror <query_json> --source-mapping .volumes/fs:nomad/fairdi/prod/fs
+    nomad client -n <api_url> -u <username> -w <password> mirror <query_json> --source-mapping <target_docker_path>:<shared_path>
 
 Here is a breakdown of the different arguments:
 
@@ -35,10 +35,10 @@ Here is a breakdown of the different arguments:
   * :code:`--source-mapping <mapping>`: The deployments use a separate folder to store
     the archive and raw data. To correctly find the data that should be
     mirrored, the absolute path on the filesystem that is shared between the
-    deployments needs to be provided. E.g. *.volumes/fs:nomad/fairdi/prod/fs*.
+    deployments needs to be provided. E.g. *.volumes/fs:/nomad/fairdi/prod/fs*.
     The first part of this mapping indicates a docker volume path
     (*.volumes/fs* in this example) that should be mapped to the second
-    filepath on the shared filesystem (*nomad/fairdi/prod/fs* in this example).
+    filepath on the shared filesystem (*/nomad/fairdi/prod/fs* in this example).
 
 Updating the AFLOW prototype information
 """"""""""""""""""""""""""""""""""""""""
@@ -53,8 +53,6 @@ The dataset should be recreated if the AFLOW dataset has been updated or if the
 symmetry matching routine used within NOMAD is updated (e.g. the symmetry
 tolerance is modified). To produce a new dataset run the following command:
 
-
-
 .. code-block:: sh
 
     nomad admin ops prototypes-update <module_path>