Adding pytorch support and switching to rllib Repeated action space

What's Changed

• WIP | adding draft of hallway documentation by @pstjohn in #7
• initial example of solving tsp with a gnn by @pstjohn in #11
• Update run_hallway.py by @pstjohn in #13
• WIP | Solve TSP for larger instances using GNN by @davebiagioni in #12
• add function for random tsp baseline by @pstjohn in #14
• Doc generation improvements and templates by @xtruan in #15
• Pin gym for rllib by @pstjohn in #18
• Test alternative RL strategies by @pstjohn in #17
• WIP | Upgrades to TSP to beat nx baseline by @davebiagioni in #16
• Refactor TSP GNN preprocessing to improve speed by @pstjohn in #19
• Add config and doc for TSP on Ray at AWS by @dmdu in #22
• still some issues, but autogenerates package documentation by @pstjohn in #23
• Random graph by @pstjohn in #20
• Repeated space masking by @pstjohn in #28
• playing around with Repeated observation space by @pstjohn in #27
• Enable the use of torch in graph-env with examples by @davebiagioni in #29

New Contributors

• @xtruan made their first contribution in #15
• @dmdu made their first contribution in #22

Full Changelog: v0.0.3...v0.0.4