Improve CRDs situation

[gabschne]

Hi,

I currently see some issues in relation to the CRDs which, if resolved, would improve the deployment and operational experience with Redpanda:

• Automatic installation, updates and uninstallation of CRDs (with the Helm Chart): Currently installing CRDs is a manual step, which means I have to watch for updates myself or create a CRON job applying them regularly.
  This is related to Redpanda-Operator: Controller should manage its own crds redpanda#9321.
  It's also worth noting that the helm chart or operator already seems to install some CRDs from the monitoring.coreos.com group, which are normally installed by the Prometheus stack. This is another inconsistency in the current deployment process.
• Seperate the Flux CRDs from the Redpanda CRDs. This point is already tracked by Provide a version of our CRDs without bundling in Flux CRDs #98, however I want to mention it here again for the sake of completeness and because I think it's a very important point. If the installation command for the CRDs is being executed in a cluster that is already managed by Flux (and therefore already as those CRDs installed), this results in errors.
• Nit in additon to the previous point: The current installation command for the CRDs in the docs uses a GitHub link which works when executing it but results in a 404 if tried to open with a browser. So if someone, following the good old "Don't just copy code from the internet and execute it" rule, tries to follow the link and wants to view the files to be deployed, the person may be confused and think the link is incorrect. It might therefore be worth it to add a small info box to the command that points out this fact, and with a link that can be opened in the browser.
  I can make that simple change in the docs and open a PR, if wantend.