Improve Error Handling and Logging

[cartermak]

This is a general task to improve error handling and messages, including handling common special cases on a per-query basis. Examples:

• Throw a TimeoutError instead of a RuntimeError for Hasura timeouts (common on large, complex queries)
• Provide a message explaining that a plan ID is not available when trying to download a plan that doesn't exist

[cartermak]

In general, I think there could be a larger task in implementing central logging in Aerie CLI that could print to stdout when used in a Python package, but have a custom stream handler implemented for CLI operation that would be able to wrap logs for pretty formatting and even support flexible log levels.

[cartermak]

Expanded idea:

The main entrypoint should catch all exceptions and gracefully exit. Because aerie-cli already has persistent file storage, we can add log files to that location and provide the full traceback there.

[cartermak]

General goals:

• Anytime Aerie-CLI throws an error that we don't handle gracefully, the full stacktrace should be save to a log file somewhere. Only a short message should be printed to the terminal, instead of the full stacktrace.
• At the top level (probably in main() of __main__.py), we should catch all exceptions such that the program always exists gracefully
• Implement logging such that we can log debug-level, info-level, etc. from all methods in Aerie-CLI. This will include defining a top-level logger object somewhere that's accessible from all the commands and AerieClient. A custom stream handler should print pretty log messages to the terminal (using the rich library). An additional Aerie-CLI global argument can configure the logging level.