Write results to cwl.output.json to bypass size restrictions

This MR modifies the way LINC collects the output in the filter_ms_group step. The output is written to cwl.output.json, which is then parsed as part of the collection of the outputs. This the loadContents field to circumvent size restrictions.

Solves #63 (closed).

steps/filter_ms_group.cwl

@@ -22,16 +22,15 @@ outputs:
     type: Directory[]
     outputBinding:
       glob: $(runtime.outdir)/cwl.output.json
-      outputEval: '$(JSON.parse(self[0].contents).selected_ms_names)'
+      outputEval: $(self.output)
   - id: selected_ms
     type: string[]
     outputBinding:
         glob: $(runtime.outdir)/cwl.output.json
-        outputEval: $(JSON.parse(self[0].contents).selected_ms)
+        outputEval: $(self.selected_ms)
 label: filter_ms_group
 
 requirements:
-  - class: InlineJavascriptRequirement
   - class: InitialWorkDirRequirement
     listing:
      - entryname: filter_ms_group.py
@@ -55,7 +54,7 @@ requirements:
         selected_ms_names = [ms_by_name[ms_name] for ms_name in selected_ms
                                                     if ms_name != 'dummy.ms']
         cwl_output  = {'selected_ms': selected_ms,
-                       'selected_ms_names': selected_ms_names}
+                       'output': selected_ms_names}
 
         with open('./cwl.output.json', 'w') as fp:
             json.dump(cwl_output, fp)