Consider Moving Saving Results Functionality to Plugin

[the-other-james]

The '--save' flag saves execution results to a file, but is not currently used/supported within plugins. I think this functionality could actually be moved to a plugin that writes the high level dictionary to a file that can then be optionally included. From an internal discussion:

Why this is better implemented in a plugin:

1. We want to keep the execution engine simple <- a simple implementation is possible, but we may want to support multiple file formats. Or making sure you don't overrun your disk quota. I would rather keep execution engine simple and implement a simple version in a plugin. Folks that just need the dumb version are all set. Folks that need more complicated logic can implement there own plugin based off of our example.
2. No every use case will want the functionality <- for example hardware may not have the resources to constantly write out to disk. Or someone running experiments with the goal of repeatability may want very verbose output.
3. The plugin architecture already exists

[asgibson]

I think OnAIR should provide a rudimentary save feature that can at least record some data/results. It could be able to be turned on or off via config file. The plugin structure looks to be more centered on AI constructs and not simple operations like saving data. However, if someone desired a more complex saving structure, there is nothing stopping them from using a plugin or adding such a feature to a plugin. The last Complex Reasoner plugin would be an ideal location for such a saver as it should have access to all previous results.

[the-other-james]

I think the csv_output plugin (https://github.com/nasa/OnAIR/blob/main/plugins/csv_output/csv_output_plugin.py) gives us the functionality we need by writing low and high-level data out to a file.

@asgibson do you object to me removing the --save flag and related functionality?