API key authentication 2.0

[guicassolato]

Placeholder for kicking off the investigation on a set of significant improvements to the API key authentication feature, aiming for:

1. Improve memory utilisation – possibly by:
   • No longer indexing all API key Secrets in memory - use a cache system instead (?)
   • Avoid multiple copies of the Secrets for each AuthConfig it matches
   • Reduce the amount of data associated to each API key secret that is stored memory
     • Exclude unnecessary fields - e.g., metadata.annotations["kubectl.kubernetes.io/last-applied-configuration"], metadata.managedFields[] - any use case for these (?)
     • Make it configurable (?)
2. Improve reconciliation performance and observability
   • Reduce time to reconcile (at scale)
   • Status reporting - none today
   • Logging - possibly too noisy today by misreporting secrets that are actually out of scope
   • Tracing and metrics - probably nothing to be done
3. Integration with external API key authentication providers and secret management solutions (e.g. Vault)
4. Better multi-cluster story