when re-run with a more restrictive set of sample data flags (--catalog, --factsets, --reports).
(Currently this fails because I had a bug in f43a2ca90.)
Also fixes a bug in catalog/utils/add-random-resource-to-catalog where we were creating a random-resource, getting a key from its properties, then associating that key with yet another random resource rather than one we first created. Over long runs of benchmark this was generating errors during parsing of the catalogs.
when re-run with a more restrictive set of sample data flags (--catalog, --factsets, --reports).
(Currently this fails because I had a bug in f43a2ca90.)
Also fixes a bug in catalog/utils/add-random-resource-to-catalog where we were creating a random-resource, getting a key from its properties, then associating that key with yet another random resource rather than one we first created. Over long runs of benchmark this was generating errors during parsing of the catalogs.