# Description

This MR adds the dexterous cube manipulation environment from
IsaacGymEnvs. The implementation is mostly based on the standard
AllegroHand environment. However, it includes the following components
from the Dextreme work:

* Randomization of mass, joint PD gains, friction, and initial state
distribution
* Tuning of RL-Games from Dextreme work
* Exponential moving average (bounded joint position) action term

However, it does the following differently since it led to better
convergence:

* Changes the way out-of-reach termination is computed. Original work
seems to do it w.r.t. the goal position, but that seems unnecessary
* Removed goal position racking reward. It was tuned too high, which
made learning difficult and is not needed

## Type of change

- New feature (non-breaking change which adds functionality)

## Screenshots

Trained with RSL-RL


https://github.com/isaac-orbit/orbit/assets/12863862/8f51e468-2d93-4520-9689-f1e8f1a898e6

## Checklist

- [x] I have run the [`pre-commit` checks](https://pre-commit.com/) with
`./orbit.sh --format`
- [x] I have made corresponding changes to the documentation
- [ ] My changes generate no new warnings
- [ ] I have added tests that prove my fix is effective or that my
feature works
- [x] I have run all the tests with `./orbit.sh --test` and they pass
- [x] I have updated the changelog and the corresponding version in the
extension's `config/extension.toml` file
- [x] I have added my name to the `CONTRIBUTORS.md` or my name already
exists there