Get logging working in ray

[eric-anderson]

• Log the incremental output from the server
• Handle errors that happen partway through processing
• Move ray logging setup into default sycamore init
• for testing, add a sycamore.shutdown() to clean up sycamore.init()
• Add support for not loading the model when we're only going to call the server
• Add assertions to code to check for user error
• Switch partitioner to using logger. instead of logging. It's better style, it turned out to not be necessary.
• Remove caching from aryn_config. It caused problems when called with different parameters If we're going to re-introduce this we can use @cache which will respect parameters.
• Make SycamorePartitionerExample notebook runnable. The example failed because of missing files/directories.

[MarkLindblad]

There is a leftover import of the old ray_logging_setup at test_partition.py:70 that is causing the integration tests to fail.