## ORCID

Directly from compressed tarball; takes about 2 hours in production:

    tar xf /srv/datasets/public_profiles_API-2.0_2017_10_json.tar.gz -O | jq -c . | grep '"person":' | time parallel -j12 --pipe --round-robin ./fatcat_import.py import-orcid -

After tuning database, `jq` CPU seems to be bottleneck, so, from pre-extracted
tarball:

    tar xf /srv/datasets/public_profiles_API-2.0_2017_10_json.tar.gz -O | jq -c . | rg '"person":' > /srv/datasets/public_profiles_1_2_json.all.json
    time parallel --bar --pipepart -j8 -a /srv/datasets/public_profiles_1_2_json.all.json ./fatcat_import.py import-orcid -

Does not work:

    ./fatcat_import.py import-orcid /data/orcid/partial/public_profiles_API-2.0_2017_10_json/3/0000-0001-5115-8623.json

Instead:

    cat /data/orcid/partial/public_profiles_API-2.0_2017_10_json/3/0000-0001-5115-8623.json | jq -c . | ./fatcat_import.py import-orcid -