conditional gfn

[josephdviviano]

Here's an example of how I envision conditioning working. We want conditioning information to be passed to the gflownet from the training loop, in the form of a tensor which is shape=[n_trajectories, n_cond_dims].

It's only working with TB now. But barring any major updates or changes, I think it will be easy enough to propagate this to the other gflownets.

Note that the conditioning is useless in my example, we should have a better use-case envisioned for the demo.