Saturation monitor addon with own output channel

As suggested in comment on !103 (closed), I think keeping 10 Hz saturation monitoring updates on an output channel preferable. It can, however, be a different output channel than the one for fast data such that it can bind to different interface. I had to update the interface between correction device and addons a bit for this:

• Addon needs to know its prefix in the schema
  • It used to only be passed to extend_device_schema (where we can add an OUTPUT_CHANNEL)
  • To actually writeChannel at runtime, we need to know the full name of the channel
• Addon needs to know train ID per input such that matching works
• Could extract from hash (and initially did), but that's cumbersome - and breaks in my testing environment where timeserver is made up, but replayed data has actual train IDs

src/calng/correction_addons/base_addon.py

@@ -25,12 +25,16 @@ class BaseCorrectionAddon:
         device)"""
         return self._device._geometry
 
-    def post_correction(self, processed_data, cell_table, pulse_table, output_hash):
+    def post_correction(
+        self, train_id, processed_data, cell_table, pulse_table, output_hash
+    ):
         """Called directly after correction has happened. Processed data will still be
         on GPU if the correction device is generally running in GPU mode."""
         pass
 
-    def post_reshape(self, reshaped_data, cell_table, pulse_table, output_hash):
+    def post_reshape(
+        self, train_id, reshaped_data, cell_table, pulse_table, output_hash
+    ):
         pass
 
     def reconfigure(self, changed_config):