Improve error handling and logging

[Navarro-Jonathan]

• Added a top level exception handler. This prints the exception type to the console and writes the full stack trace to a log file
• Added log files into the persistent aerie-cli directory. A new log file is created whenever an aerie-cli command is run. It is set to keep the most recent 25 log files at a time.
• Replaced typer.echo calls with logging.info
• Added a file handler and console handler to the logger. The console logger will only output the logs at the info level, but the file handler will output logs at the debug level.
• Added debug logs to the select_from_list prompts
• Added a --debug/-d CLI option to print all debug-level logs to stdout

Closes #3

[cartermak]

One other general question: does caplog only return logs generated by/during that specific test case, or does it contain all logs?

[cartermak]

Suggest centralizing this logic in src/utils/logger.py. Ideally, that module can have a function like set_log_level which is the callback for the --debug option.

[cartermak]

I know I went back and forth on this, but I decided I like the ergonomics better if we configure a specific child logger in /src/utils/logger.py. Then, it's clearer in the code that the logger we import has been specifically configured for this application. For example:

from aerie_cli.utils.logger import aerie_cli_logger as logger

...

logger.info("Whatever")

[cartermak]

I know this isn't your code, but I vote we tidy this up while we're at it:

    goal_ids = []
    for goal in clear_goals:
        goal_ids.append(goal["goal"]["id"])

    logging.info(f"Deleting goals for Plan ID {plan}: {' '.join([str(i) for i in goal_ids])}")

[cartermak]

This feels like values and logic that would be more cleanly tracked in a PersistentLogs class. For the while loop specifically, we could also solve this by appending a UUID, which will avoid the race condition where two Aerie-CLI processes both think they've found an available log path before creating it.

[cartermak]

Same comment as above -- this probably ought to be a PersistentLogs.clear_old_log_files method

[cartermak]

Do we need to name and track GLOBAL_LOGGER?

Suggested change

	GLOBAL_LOGGER = logging.getLogger(__name__)
	GLOBAL_LOGGER.propagate = True # we must set this to see stdout in caplog
	logging.getLogger().propagate = True # we must set this to see stdout in caplog

[cartermak]

I think we can define our own fixture that takes in caplog and sets the loglevel once. Then, we can just take in that fixture in each of our tests.

[cartermak]

Suggested change

	logging.debug(traceback.format_exc())
	logging.debug(traceback.format_exc())
	sys.exit(-1)