live: Add log_artifact

[daavoo]

Per treeverse/dvclive#464

[github-actions]

Link Check Report

There were no links to check!

[dberenbaum]

A few places we should think about adding it:

• api-reference/live#methods (this one is a blocker)
• get-started
• how-it-works#track-the-results

For the last one, it might be helpful to explain a bit of how to use DVC. For example:

When using Live.log_artifact("model.pt"), DVCLive will cache
the model.pt file with DVC to avoid tracking large artifacts in Git. It will
generate a model.pt.dvc metadata file, which you should track in Git.
You can retrieve the artifact from the Git commit.

[daavoo]

A few places we should think about adding it:

I have left the get-started out for now. I think will make more sense to add it there after treeverse/dvclive#465

[dberenbaum]

Thanks for including my suggestion, but I think it feels a little disconnected. Feel free to drop it and we can do a follow-up PR if that's easier.

I think we should rework this whole section. Proposed text:

# Track the results

DVCLive expects each run to be tracked by Git, so it will save each run to the
same path and overwrite the results each time. Include
[`save_dvc_exp=True`](/doc/dvclive/api-reference/live#parameters) to 
auto-track in Git as a <abbr>DVC experiment</abbr>. DVC experiments
are Git commits that DVC can find but that don't clutter your Git history 
or create extra branches in your repo.

### Track artifacts

Models and data are often large and aren't easily tracked in Git. 
`Live.log_artifact("model.pt")` will [cache](/doc/start/data-management/data-versioning)
the `model.pt` file with DVC and make Git ignore it. It will generate a `model.pt.dvc`
metadata file, which can be tracked in Git and becomes part of the experiment.
You can [retrieve](/doc/start/data-management/data-versioning#retrieving) the
versioned artifact from the Git commit.

### Run experiments with `dvc exp run`

You may [run experiments](/doc/user-guide/experiment-management/running-experiments)
using DVC <abbr>pipelines</abbr>. Once you setup your pipeline, you can run it with `dvc exp run`.
This will track the inputs and outputs of your code, cache them so you never waste time repeating 
steps, and enable other features like queuing, hyperparameter tuning, and
grid searches.

[dberenbaum]

Related: treeverse/dvclive#468 (comment) cc @shcheklein