Fix certificate hashes re-computation in aggregator

[jpraynaud]

Why

Following the merging of the PR #1333, we have run the certificate hashes re-computation runbook, we have faced multiple difficulties we'd like to fix.

What

The migration was not planned and thus not run locally previously to the merge, which gave only a very short delay to run the aforementioned migration. In order to limit the human factor we need to create a manual-deployment tag which will mark the PR needing a manual operation after merging, and a "Team Golden Rules" to which all the teammates agree.

Also the migration process needs to be more robust and fixed accordingly

How

• Activate a manual-deployment label in GitHub issues
• Redact team golden rules in the Project Charter wiki page
• Fix certificate hashes re-computation tool in aggregator:
  • Maximum variables in SQL query reached with batch insert of new certificates
  • Tolerate rec-computation where not all the certificates change (e.g. genesis certificate could not have their hash modified)
  • Remove unnecessarily required configuration parameters to run the command (keep only the database store directory)
  • Verify date format discrepancy on all networks (see InvalidData("Could not turn string '2023-05-09 13:11:15' to rfc3339 Datetime. Error: premature end of input") errors on testing-preview)
• Fix runbook with aggregator being fully stopped during migration process