Needs to save periodically: run.sh runs for the full run ATM

[yarikoptic]

Found /home/dandi/cronlib/dandisets-healthstatus/run.sh running with last commit in Nov 2023! And our last_run also growing slowly through the time:

dandi@drogon:~/cronlib/dandisets-healthstatus$ git grep 'last_run:' results | head
results/000003/status.yaml:last_run: '2024-01-03T12:21:19.322063-05:00'
results/000004/status.yaml:last_run: '2024-01-03T12:21:19.479233-05:00'
results/000005/status.yaml:last_run: '2024-01-03T12:21:19.640660-05:00'
results/000006/status.yaml:last_run: '2024-01-03T12:21:19.894148-05:00'
results/000007/status.yaml:last_run: '2024-01-03T12:21:20.025406-05:00'
results/000008/status.yaml:last_run: '2024-01-03T12:21:20.397709-05:00'
results/000009/status.yaml:last_run: '2024-01-03T12:21:20.622131-05:00'
results/000010/status.yaml:last_run: '2024-01-03T12:21:20.983880-05:00'
results/000011/status.yaml:last_run: '2024-01-03T12:21:21.149759-05:00'
results/000012/status.yaml:last_run: '2024-01-03T12:21:21.436046-05:00'
dandi@drogon:~/cronlib/dandisets-healthstatus$ git grep 'last_run:' results | tail -n 20
results/000624/status.yaml:last_run: '2024-02-26T01:09:37.332082-05:00'
results/000625/status.yaml:last_run: '2024-02-26T04:10:08.748374-05:00'
results/000626/status.yaml:last_run: '2024-02-26T05:09:45.317464-05:00'
results/000628/status.yaml:last_run: null
results/000630/status.yaml:last_run: null
results/000631/status.yaml:last_run: null
results/000632/status.yaml:last_run: null
results/000633/status.yaml:last_run: null
results/000634/status.yaml:last_run: null
results/000635/status.yaml:last_run: null
results/000636/status.yaml:last_run: null
results/000637/status.yaml:last_run: null
results/000640/status.yaml:last_run: null
results/000678/status.yaml:last_run: null
results/000680/status.yaml:last_run: null
results/000683/status.yaml:last_run: null
results/000687/status.yaml:last_run: null
results/000691/status.yaml:last_run: null
results/000692/status.yaml:last_run: null
results/000696/status.yaml:last_run: null

and I thought that we were doing 1 (or some N) file from each dandiset sweep in a loop with save... would need to check after reboot. filing to not forget