Change `logging.loggers` config from Array to Map

[joshdover]

Follow up from discussion here: #68704 (comment)

We currently expose our logging context configuration via an array format, where each logger has context key that is used to identify the logger:

loggers:
  - context: foo.bar
    level: 'info'
    appenders: 
      - file

We should change this to a Map-based format where the key is the context and the values are an object containing the level and appenders.

loggers:
  foo.bar:
    level: 'info'
    appenders: 
      - file

Elasticsearch supports this same format, so we will still be consistent with ES here. Luckily these new config keys are still experimental, so we can change these before 8.0

[elasticmachine]

Pinging @elastic/kibana-platform (Team:Platform)

[pgayvallet]

Note that the way we are reading/merging the config from file relies on _.set, meaning that atm, the dotted properties are expanded.

kibana/src/core/server/config/read_config.ts

Lines 39 to 49 in 415e3bc

	function merge(target: Record<string, any>, value: any, key?: string) {
	if ((isPlainObject(value) || Array.isArray(value)) && Object.keys(value).length > 0) {
	for (const [subKey, subVal] of Object.entries(value)) {
	merge(target, subVal, key ? `${key}.${subKey}` : subKey);
	}
	} else if (key !== undefined) {
	set(target, key, typeof value === 'string' ? replaceEnvVarRefs(value) : value);
	}
	return target;
	}

loggers:
  foo:
    level: 'info'
    appenders: 
      - file
  foo.bar:
    level: 'error'
    appenders: 
      - file

would become

{
  loggers: {
    foo: {
      level: 'info',
      appenders: ['file'],
      bar: {
        level: 'info',
        appenders: ['file'],
      },
    },
  },
};

[joshdover]

Deciding not to do this now.