Move some documents to a "Main Concepts" section, normalized filenames to only use underscores and simplified some document headings. (#755)

Author: ernestum

docs/getting-started/first-steps.rst renamed to docs/getting-started/first_steps.rst

@@ -43,12 +43,22 @@ If you use ``imitation`` in your research project, please cite our paper to help
    :caption: Getting Started
    :hidden:
 
+   getting-started/what_is_imitation
    getting-started/installation
-   getting-started/what-is-imitation
-   getting-started/variable-horizon
-   getting-started/first-steps
+   getting-started/first_steps
    getting-started/cli
 
+.. toctree::
+    :maxdepth: 2
+    :caption: Main Concepts
+    :hidden:
+
+    main-concepts/experts
+    main-concepts/trajectories
+    main-concepts/reward_networks
+    main-concepts/variable_horizon
+
+
 .. toctree::
    :maxdepth: 2
    :caption: Algorithms
@@ -77,15 +87,6 @@ If you use ``imitation`` in your research project, please cite our paper to help
    tutorials/7_train_density
    tutorials/8_train_custom_env
    tutorials/9_compare_baselines
-   tutorials/trajectories
-   tutorials/reward_networks
-
-.. toctree::
-   :maxdepth: 2
-   :caption: Experts
-   :hidden:
-
-   experts/loading-experts
 
 API Reference
 ~~~~~~~~~~~~~

docs/getting-started/what-is-imitation.rst renamed to docs/getting-started/what_is_imitation.rst

@@ -1,6 +1,6 @@
-===============
-Loading Experts
-===============
+=======
+Experts
+=======
 
 The algorithms in the imitation library are all about learning from some kind of
 expert.

docs/index.rst

@@ -1,6 +1,6 @@
-===================
-Handle Trajectories
-===================
+============
+Trajectories
+============
 
 For imitation learning we need trajectories.
 Trajectories are sequences of observations and actions and sometimes rewards, which are generated by an agent

docs/experts/loading-experts.rst renamed to docs/main-concepts/experts.rst

@@ -3,8 +3,8 @@
 Limitations on Horizon Length
 ================================================
 
-Variable Horizon Environments Considered Harmful
-================================================
+.. warning:: Variable Horizon Environments Considered Harmful
+
 
 Reinforcement learning (RL) algorithms are commonly trained and evaluated in *variable horizon* environments.
 In these environments, the episode ends when some termination condition is reached (rather than after a fixed number of steps).