As per recommendations from AWS at Fluent Con, determine the `Mem_Buf_Limit` to best use for each input configuration: https://docs.fluentbit.io/manual/pipeline/inputs/tail
We probably want to get this from the resource limits on the container (if defined) via the DownwardAPI route - provide as an env var and use that (or default it if not defined).
Other recommendations are to enable the filesystem buffering as well, limited by any limits we can discover too. We should make sure the metrics for all this are also enabled.
|For Gerrit Dashboard: K8S-2172|
|158650,3||K8S-2172: Memory limits for tail logging||master||couchbase-operator||Status: ABANDONED||0||+1|
|168381,26||K8S-2172: Added mem buf limits for inputs||main||couchbase-fluent-bit||Status: MERGED||+2||+1|
|170797,9||K8S-2172: Updated mem buf limits env var||master||couchbase-operator||Status: MERGED||+2||+1|