Allow logging configuration from configuration yaml file

This will allow proper logging configuration for the services which are currently running in the dynamic infrastructure. Their logs are current written in the wrong elasticsearch indices.

The existing behavior is kept as is so we can focus on changing the current logging configuration deployment incrementally.

And then, some json logging configuration file could look like:

---
version: 1

handlers:
  console:
    class: logging.StreamHandler
    formatter: json
    stream: ext://sys.stdout

formatters:
  json:
    class: pythonjsonlogger.jsonlogger.JsonFormatter
    # python-json-logger parses the format argument to get the variables it actually expands into the json
    format: "%(asctime)s:%(threadName)s:%(pathname)s:%(lineno)s:%(funcName)s:%(task_name)s:%(task_id)s:%(name)s:%(levelname)s:%(message)s"

loggers:
  celery:
    level: INFO
  amqp:
    level: WARNING
  urllib3:
    level: WARNING
  azure.core.pipeline.policies.http_logging_policy:
    level: WARNING
  swh: {}
  celery.task: {}
  
root:
  level: <<loglevel>>
  handlers:
  - console

While a systemd logging config matching the current setup for celery would look like:

---
version: 1

handlers:
  console:
    class: logging.StreamHandler
    formatter: task
    stream: ext://sys.stdout
  systemd:
    class: swh.core.logger.JournalHandler
    formatter: task

formatters:
  task:
    (): celery.app.log.TaskFormatter
    fmt: "[%(asctime)s: %(levelname)s/%(processName)s] %(task_name)s[%(task_id)s]: %(message)s"
    use_color: false

loggers:
  celery:
    level: INFO
  amqp:
    level: WARNING
  urllib3:
    level: WARNING
  azure.core.pipeline.policies.http_logging_policy:
    level: WARNING
  swh: {}
  celery.task: {}

root:
  level: DEBUG
  handlers:
  - console
  - systemd

Note: the task_name and task_id qualifiers are only available within celery, when using the TaskFormatter

Ref. swh/infra/sysadm-environment#4524 (closed) Depends on swh-core!335 (merged)