Thinking about STIPS configuration parameters

[york-stsci]

Currently, STIPS has a bunch of configuration parameters, and all of them are treated the same way. They have a default value kept inside STIPS itself, can be given a default value by creating a new configuration file and telling STIPS to look at it, and can be set at run-time by providing values when creating top-level objects (SceneModule and ObservationModule).

For some configurations (e.g. log level, data output location, whether to use parallel processing, PSF oversampling) this makes a lot of sense, because a single observation (or scene) should probably use the same values throughout. However, for some configuration parameters (e.g. random number seed) it probably makes more sense to let each detector in an observation have a separate value (although, especially for testing/reproducibility purposes, having that separate value be generated in a known way does likely make sense.