feat: migrate to dynaconf to handle multi-source configuration

[xoxys]

BREAKING CHANGE: With the migration to a new configuration framework, we have also restructured the configuration file and environment variables. Please check the documentation before you upgrade.

New config layout:

---
# Default is the current working directory.
base_dir:

role:
  # Default is the basename of 'role_name'.
  name:
  # Auto-detect if the given directory is a role, can be disabled
  # to parse loose files instead.
  autodetct: True

# Don't write anything to file system
dry_run: False

exclude_files: []
# Examples
# exclude_files:
#   - molecule/
#   - files/**/*.py

# Exclude tags from automatic detection. Configured tags are only skipped
# if the tag is not used in an annotation.
exclude_tags: []

logging:
  # Possible options debug | info | warning | error | critical
  level: "warning"
  # Json logging can be enabled if a parsable output is required
  json: False

template:
  name: readme
  # Default is built-in templates directory.
  src:

  options:
    # Configures whether to tabulate variables in the output. When set to `True`,
    # variables will be displayed in a tabular format intsead of plain marktdown sections.
    # NOTE: This option does not support rendering multiline code blocks.
    tabulate_vars: False

renderer:
  # By default, double spaces, spaces before and after line breaks or tab characters, etc.
  # are automatically removed before the template is rendered. As a result, indenting
  # with spaces does not work. If you want to use spaces to indent text, you must disable
  # this option.
  autotrim: True
  # Load custom header from given file and append template output to it before write.
  include_header: ""
  # Path to write rendered template file. Default is the current working directory.
  dest:
  # Don't ask to overwrite if output file exists.
  force_overwrite: False