Added more robust archive file corruption handling

This PR will add an additional check when reading archive hdf5 files so that each group is read once to check that there are no runtime errors due to file corruption in the group

gwsumm/archive.py

@@ -200,6 +200,18 @@ def read_data_archive(sourcefile, rm_source_on_fail=True):
 
     with File(sourcefile, 'r') as h5file:
 
+        # Make sure that each group is not corrupted by trying to read the data
+        for group in h5file.keys():
+            try:
+                dataset = h5file.get(group, {})
+            except RuntimeError as exc:
+                if not rm_source_on_fail:
+                    raise
+                warnings.warn(f"failed to read {sourcefile} group {group} "
+                              f"[{exc}], removing...")
+                os.remove(sourcefile)
+                return
+
         # -- channels ---------------------------
 
         try: