Change default logging.yml to use RichHandler 1: initial setup

[antonymilne]

🥳 FUN TICKET 🥳 Prerequisites: #1470 and #1469. Based off point 2 of #1464

• add rich as a core kedro dependency
• add the following handler to framework-side logging.yml

rich:
    class: rich.logging.RichHandlers

• for any loggers that have console listed under the handlers (probably just root), swap console for rich. Leave the console handlers still defined though
• make the same change in the project-side logging.yml (+ starters + e2e test)
• play around with the different options that are possible under rich - read https://rich.readthedocs.io/en/stable/reference/logging.html and https://rich.readthedocs.io/en/stable/logging.html to see what these are
  • there should be no need to set level but check this works right (should show messages >= INFO for kedro, >= WARNING for all other loggers)
  • generally play around with the options - probably the defaults are suitable for the most part but feel free to be creative, see what looks nice with a kedro run, e.g. do we want to set log_time_format: [%X]?
  • don't change any options for tracebacks; that is for Change default logging.yml to use RichHandler 2: fancy tracebacks #1513
  • don't set markup: True. Since we're modifying the root logger that would be too risky as other libraries might not be rich-aware and use [] that breaks stuff. We'll do markup for kedro logging in another ticket
• test thoroughly to check that kedro run looks nice and what happens if you emit log messages from outside the kedro logger
• test what happens if you do session.run in ipython and jupyter
• add docs to explain how a user could switch back from rich to console by modifying their project-side logging.yml As in Write documentation on how to make a custom logging.yml in your project #1474, this should be as easy as possible

[datajoely]

Could we make the markup=False configurable?

[antonymilne]

It will be - I just added more detail to the ticket to explain this. I'm very keen on using markup=True where we can in a follow-up ticket, just don't want to apply it to all loggers in the first step here.