Autoregressive model on resting state time series benchmark

Based on https://github.com/FrancoisPgm/fmri-autoreg

I have my own fork as lots of changes to optimise for large n dataset.

• All inputs (i.e. building blocks from other sources) are located in inputs/.
• All custom code is located in code/.

Python version 3.10.x. The reference model we are using is code in pytorch 1.13.1.