Declarative log level

[rzetelskik]

What should the feature do?

Loglevel of particular components should be configurable through ScyllaOperatorConfig.

What is the use case behind this feature?

Allows for finer granularity and is a step towards having a centralised config.

Anything else we need to know?

For backwards compatibility, the flag provided to the binary should take precedence.

Duplicate of #836.

/lifecycle frozen
/priority important-soon

[mflendrich]

Merged issues for scylla cluster loglevel & operator machinery loglevel into one.

From the user perspective, having a consistent loglevel experience justifies handling this holistically. Maybe the loglevel setting can/should be hierarchical - to be decided when implementing.