Crash on resign when preview/main collections are missing

[leplatrem]

echo '{"data": {"status": "to-resign"}}' | http PATCH $SERVER/buckets/security-state-staging/collections/onecrl --auth admin:$ADMIN_PASS
HTTP/1.1 500 Internal Server Error
Access-Control-Allow-Origin: *
Access-Control-Expose-Headers: Retry-After, Alert, Content-Type, Content-Length, Backoff
Alt-Svc: clear
Content-Length: 177
Content-Security-Policy: default-src 'none'; frame-ancestors 'none'; base-uri 'none';
Content-Type: application/json
Date: Tue, 09 Jul 2024 16:43:48 GMT
Server: nginx
Strict-Transport-Security: max-age=31536000
Via: 1.1 google
X-Content-Type-Options: nosniff

{
    "code": 500,
    "errno": 999,
    "error": "Internal Server Error",
    "info": "https://github.com/Kinto/kinto/issues/",
    "message": "A programmatic error occured, developers have been informed."
}

• ⚠️ No Sentry entry for this 500
• ⚠️ The refresh_signature cronjob should also hit this, and we were not aware of it
• ℹ️ the preview and main collections of most collections in the security-state-staging bucket don't exist (eg. onecrl). They are well created though when calling echo '{"data": {"status": "to-sign"}}' | http PATCH $SERVER/buckets/security-state-staging/collections/{cid}

[leplatrem]

Explanation:

• users never published anything on these collections
• collections are not listed on https://remote-settings-dev.allizom.org/v1/buckets/monitor/collections/changes/records
• the refresh signature cronjob does not hit them because that its source of truth