Add Data Version Control (DVC) for tracking mapping experiments

[kursatyurt]

ASTE is used to measure numerical and performance metrics of preCICE.

One of the goals is having a reproducible experiment environment and allow everybody to share their results. It can be also used in development phase to measure how the changes affect metrics.

DVC is open-source system for data management. It is mainly designed for Machine Learning, however applicable to cases where huge amount of data plays role.

https://github.com/kursatyurt/aste/tree/dvc_v2 has a reference usage of DVC for ASTE experiments and looking for your feedback to how to introduce DVC to ASTE pipeline.

The DVC pipeline which is similar to github action pipelines can be seen in https://github.com/kursatyurt/aste/blob/dvc_v2/dvc.yaml

We are looking for your feedback 🥰

[fsimonis]

The plotting capabilities in DVC are really minimal.
One needs to specify X and Y for each series.
So we need a custom config for each experiment based on the test series.

For each plot, we need:

• a dvc configuration specifying mesh width (X) and series (Ys)
• a CSV with mesh width;NN;NP;TSP

We can easily generate these from the result file after gathering results from the test runs.