Scaffolding for config file handling and model selection

[drewoldag]

This PR represents the scaffolding work for two aspects of fibad:

1. Basic support for reading and loading the default and the user's defined runtime configuration file for use within fibad.
2. Basic support for dynamically loading models, either from a registry of built in models or via a user defined module spec, specified in the user's runtime configuration file.

To use a built in model from the registry a user specifies in the .toml config file the following:

[train]
model_name = "name_of_built_in_model_class"

To specify a model outside of fibad, the users .toml config would include something like the following example. Basically what the user would write in a from ... import ... statement.

[train]
model_class = "package.submodule.subsubmodule.ModelClass" 

Included here is an implementation of PyTorch's "Getting Started" CNN for classifying images in the CIFAR10 dataset - note that this is incomplete, and meant to be a way to test the model registration process, and will be a testbed for future work.

[github-actions]

Before [0cd1e6c]	After [174e587]	Ratio	Benchmark (Parameter)
2.23±1s	3.42±0.9s	~1.53	benchmarks.time_computation
856	3.82k	4.47	benchmarks.mem_list

Click here to view all benchmarks.

[codecov]

Welcome to Codecov 🎉

Once you merge this PR into your default branch, you're all set! Codecov will compare coverage reports and display results in all future pull requests.

Thanks for integrating Codecov - We've got you covered ☂️

[mtauraso]

LGTM. Good to add tests, but I'm in favor of this getting merged even as-is.