Iterating on Simulator Design for Sampling

[sreekaroo]

Overview

Focused on high-level design for the simulator. Modularized parts of the simulator so we can iterate on designs more easily. This draft PR is mostly just to share code easily. The goal was to create a simple data ingestor and sampler that could sample by letter from multiple data sources. View sim_test.py for examples of use.

I think the replay session logic will fit into this design without any change as well. The sampler would just be an iterator, that I implemented in the previous design. The DataEngine would load in-session data instead, which is also already implemented.

Ticket

Link a pivotal ticket here

Contributions

• Created interfaces for DataEngine, Sampler, MetricReferee, ModelHandler, StateManager
• Implemented RawDataEngine and SimpleLetterSampler, which can successfully load in multiple data sources, organize into queryable schema, and return eeg samples that can be fed into SignalModel
• Implemented draft StateManager that handles a lot of the updating logic from inquiry to inquiry

Started but unfinished

• The simulator run loop needs a little more work, view in sim_sampler.py
• MetricReferee, ModelHandler, all have unfinished implementations
• StateManager is a bit naive right now. There is also an issue with how I'm evaluating evidence because the simulator is always making a decision; albeit, mostly wrong ones.

Test

-No unit testing yet, just ran script to test sampling and data loading in sim_test.py

Documentation

• Are documentation updates required? In-line, README, or documentation? Verify the updates you did here.

Changelog

• Is the CHANGELOG.md updated with your detailed changes?