Introduce min_instances parameter

[Yusuf-Uzun]

Introduce the min_instances parameter. Adjust the max_instances parameter to reflect that it now also leverages memory.

[juergen-walter]

Suggested change

Specifies the minimum number Fluentd instances which are always provisioned regardless of auto-scaling. Utilized to guarantee a minimum number of running instances, support predictable workloads, and mitigate autoscaling lag during sudden ingestion bursts. Must be between `2` and `10` and less than or equals to `max_instances`. Default is `2`. This parameter has no effect on the *dev* plan, which is limited to a single instance. If `max_instances` is smaller than `min_instances`, both parameters will be set to the value of `max_instances` (= no autoscaling).

Specifies the minimum number ingest instances which are always provisioned regardless of auto-scaling. Utilized to guarantee a minimum number of running instances, support predictable workloads, and mitigate autoscaling lag during sudden ingestion bursts. Must be between `2` and `10` and less than or equals to `max_instances`. Default is `2`. This parameter has no effect on the *dev* plan, which is limited to a single instance. If `max_instances` is smaller than `min_instances`, both parameters will be set to the value of `max_instances` (= no autoscaling).

[juergen-walter]

This has cost implications. Makes only sense if workload is alternating so that auto-scaling cannot cover it.

[Yusuf-Uzun]

This is an intended future change that should be documented after its release to live landscapes