Underscores.me CLI Commands

I use underscores.me as a starter template for any new theme I create. Previously I just used dart-sass to compile my scss files into css. However, I'm starting a new project and would like to use the package.json file included with underscores (Sassified and with Woo Commerce), but I'm a bit confused about the included CLI commands that are in the package.json.

The following scripts are taken from the generated package.json file:

"scripts": {
    "watch": "node-sass sass/ -o ./ --source-map true --output-style expanded --indent-type tab --indent-width 1 -w",
    "compile:css": "node-sass sass/ -o ./ && stylelint '*.css' --fix || true && stylelint '*.css' --fix",
    "compile:rtl": "rtlcss style.css style-rtl.css",
    "lint:scss": "wp-scripts lint-style 'sass/**/*.scss'",
    "lint:js": "wp-scripts lint-js 'js/*.js'",
    "bundle": "dir-archiver --src . --dest ../_s.zip --exclude .DS_Store .stylelintrc.json .eslintrc .git .gitattributes .github .gitignore README.md composer.json composer.lock node_modules vendor package-lock.json package.json .travis.yml phpcs.xml.dist sass style.css.map yarn.lock"
  }

Here are my questions:

1. What is the main difference between "watch" and "compile:css"? If one watches Sass files for changes and then compiles to CSS, how is this any different from compile:css? Should they both be running?
2. Are these all scripts that should be running simultaneously during dev/coding? Or, are some supposed to be run only at specific times, such as "bundle" when I need to recompile certain files like JavaScript to compile a production version?
3. If these are not meant to all be running simultaneously, such as the "bundle" command, when is the ideal time to run this script and once it's run should I stop it or simply let it keep running in the background?
4. Would I only use "compile:rtl" if using languages written from right-to-left?
5. Since the linting commands check for errors in SCSS and JS, should those be run simultaneously during development as well, or only when you go to check work prior to deploying to production?

So assuming I do not need to compile:rtl, how would one use the remaining 5 scripts most effectively and efficiently?