Updated dependencies and patched vulnerabilities

No more patches for the foreseeable future

#### Refactorings

* (policy): streamline TorchPolicy and free functions

#### Others
* (deps): support ray-1.5.1
* substitute nnrl for raylab.torch
* (deps): add nnrl

Update package dependencies

#### Fixes

* (cli): get progress column indexes only if present in data

#### Fixes

* (deps): lock ray version to <1.3

#### Fixes

* (cli): downgrade bokeh to 2.2.2
* (cli): use future annotation style

#### Refactorings
* run isort on project
* (utils): organize exp data loading and make test

#### Docs
* (examples): update PG cartpole defaults

#### Others
* (deps): bump cachetools from 4.2.1 to 4.2.2
* (deps): bump pytorch-lightning from 1.2.9 to 1.2.10
* (deps): bump dataclasses-json from 0.5.2 to 0.5.3
* (deps): bump pytorch-lightning from 1.2.8 to 1.2.9
* (cli): annotate dashboard functions
* (cli): use subprocess.run to launch streamlit
* (deps): bump wandb from 0.10.26 to 0.10.27
* (deps): bump pytorch-lightning from 1.1.5 to 1.2.8
* (deps): run poetry lock
* (deps): bump pytorch-lightning from 1.2.7 to 1.2.8
* (deps): bump wandb from 0.10.25 to 0.10.26
* (deps): bump dm-tree from 0.1.5 to 0.1.6

#### New Features

* (utils): suppress PytorchLightning outputs intelligently

#### Others
* (utils): fix spelling error
* (examples): update ACKTR on pybullet cheetah
* (agents): remove unused configs in ACKTR
* ignore deprecation warnings from wandb

#### Others

* (examples): update ACKTR on pybullet cheetah
* (agents): remove unused configs in ACKTR

#### Fixes

* (options): update rllib config info for ray==1.2.0

#### Refactorings
* (policy): update lightning docs, signatures and annotations

#### Others
* set default framework option to 'torch' for trainers and policies
* (tests): remove redundant parenthesis
* (agents): pass framework as 'torch' in policy rollout test
* (policy): change model parameters to enable checkpoint tests

#### Others

* (torch): encase KFAC monitoring in a try-catch clause