object store connection failures

When running `integrate` workflow on the 89 tiles (in parallel), connection to nrs.objectstore.bc.ca can fail on about 10% of the tiles.
Failure is to every connection attempt from a given runner. Rerunning the workflow (failed jobs only) will generally complete.

Issue is not with ogr or with a delay to network initialization for the given runner. The runner cannot connect via curl or after a delay - the runner's ip seems to be blocked/restricted by the object store.

Presumably this is because of the volume of concurrent requests - 89 parallel jobs plus ogr's multithreading requests for parquet files.

There isn't much need for such extreme parallelization - whether the job completes in 10min vs 2hr is unimportant, hopefully restricting the concurrent runners fixes the issue.

Start by locating the `integrate` workflow and its parallel job configuration. Run the workflow or inspect its failed jobs with reduced concurrency, then confirm that all 89 tiles complete without object-store connection failures.

Written by the indexing model from the issue text.