BinaryDiscreteTensorSpec vs DiscreteTensorSpec

[matteobettini]

I recently saw that the default done_spec is

 self.output_spec["done"] = DiscreteTensorSpec(
            n=2, shape=(*self.batch_size, 1), dtype=torch.bool, device=self.device
        )

Is there a reason that this is not a BinaryDiscreteTensorSpec?
Should we make BinaryDiscreteTensorSpec just call DiscreteTensorSpec with n=2 to simplify the codebase?

[vmoens]

I'm in for simplication yep.
We could inherit from DiscreteTensorSpec and put a warning saying that it will be deprecated some time in the future, wdyt?

[matteobettini]

Yep, even without deprecating it, but just using all the logic from DiscreteTensorSpec