improve helm chart

[szaimen]

• make mail config configurable via helm chart helm chart - allow to configure additional values #3756
• make app include list configurable via helm chart helm chart - allow to configure additional values #3756
• allow to add subscription key via env helm chart - allow to configure additional values #3756
• make storage customization also configurable via helm chart not sure what is meant here but was decided against
• load balancer possibility to start Apache first (check +feedback if feasible); maybe use Put STUN/Turn server on 443 TCP/UDP (Add SSLH) or use caddy? #3576? tried but not feasible -> implement logic to get ip-address and adjust firewall config automatically
• nodes do not come up again, currently no re-scheduling (check, possibly longhorn config topic); check if using ReadWriteMany fixes this recommended to set some longhorn settings differently...
• Pin included apps to specific versions? Hopefully not necessary due to make sure that notify-push is executable #3707
• apache does not restart together with nextcloud and does not get its new ip-address on new deploy... tested on our own and seems to work

[st3iny]

I developed a unified load balancer POC and thoroughly discussed it with Simon. Aside from the fact that there were problems with Talk, we came to the conclusion to drop this idea.

Maintaining a dedicated load balancer container just for Helm is not feasible as we won't be able to guarantee stability. It won't be a part of the regular dockerized AIO setup and thus not much testing would be done on it.

Additionally, it might be tedious to maintain as my implementation turned out to be very brittle and many things had to be fixed by trial and error.

So the recommended solution from our side is indeed that the customer implements a way to retreive the ip-address that got allocated by kubernetes and autommatically adjust the firewall rules based on that. (Could potentially be done via a script using kubectl).

[szaimen]

Most things get added in #3756