add_job_records: improve error handling

Problem: The add_job_records() function does not handle the case where
a ResourceSet object cannot be created out of the R fetched for each
job.

Add a try-except block for this, and just skip adding the record in the
case where the creation of the ResourceSet object does not work since
it will not contribute to an association's job usage or fair-share
value.

src/bindings/python/fluxacct/accounting/job_archive_interface.py

@@ -123,7 +123,13 @@ def add_job_records(rows):
     job_records = []
 
     for row in rows:
-        rset = ResourceSet(row[6])  # fetch R
+        try:
+            # attempt to create a ResourceSet from R
+            rset = ResourceSet(row[6])
+            nnodes = rset.nnodes
+        except (ValueError, TypeError):
+            # can't convert R to a ResourceSet object; skip it
+            continue
 
         job_record = JobRecord(
             row[0],  # userid
@@ -132,7 +138,7 @@ def add_job_records(rows):
             row[2],  # t_submit
             row[3],  # t_run
             row[4],  # t_inactive
-            rset.nnodes,  # nnodes
+            nnodes,  # nnodes
             row[6],  # resources
         )
         job_records.append(job_record)