Adjust one of the Page Label unit-tests to use a PDF file where the "St" entry is both present and non-default (i.e. greater than one)

I just realized that none of our current unit-tests cover this particular part of the Page Label parsing code, hence this patch adjusts an existing test PDF to include a "St" entry in the Page Label dictionary.

test/unit/api_spec.js

@@ -515,7 +515,7 @@ describe('api', function() {
         expect(pageLabels[0]).toEqual(['i', 'ii', 'iii', '1']);
         expect(pageLabels[1]).toEqual(['Front Page1']);
         expect(pageLabels[2]).toEqual(['1', '2']);
-        expect(pageLabels[3]).toEqual(['X1']);
+        expect(pageLabels[3]).toEqual(['X3']);
 
         loadingTask0.destroy();
         loadingTask1.destroy();