feature(nyz): add stochastic dueling network

[PaParaZz1]

Description

Stochastic Dueling Network proposed in ACER

Related Issue

TODO

Check List

• merge the latest version source branch/repo, and resolve all the conflicts
• pass style check
• pass all the tests

[codecov]

Codecov Report

Merging #234 (0312390) into main (5dc47cc) will decrease coverage by 0.09%.
The diff coverage is 96.00%.

@@            Coverage Diff             @@
##             main     #234      +/-   ##
==========================================
- Coverage   86.38%   86.29%   -0.10%     
==========================================
  Files         461      464       +3     
  Lines       35167    36351    +1184     
==========================================
+ Hits        30379    31369     +990     
- Misses       4788     4982     +194     

Flag	Coverage Δ
unittests	86.29% <96.00%> (-0.10%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
ding/model/common/head.py	99.03% <94.11%> (-0.97%)	⬇️
ding/model/common/tests/test_head.py	100.00% <100.00%> (ø)
ding/model/template/q_learning.py	86.51% <0.00%> (-9.73%)	⬇️
ding/model/template/atoc.py	94.21% <0.00%> (-1.06%)	⬇️
ding/envs/env_manager/subprocess_env_manager.py	79.91% <0.00%> (-0.22%)	⬇️
ding/policy/__init__.py	100.00% <0.00%> (ø)
ding/torch_utils/network/__init__.py	100.00% <0.00%> (ø)
ding/torch_utils/network/gtrxl.py	97.20% <0.00%> (ø)
ding/torch_utils/network/tests/test_gtrxl.py	100.00% <0.00%> (ø)
ding/policy/r2d2_gtrxl.py	17.81% <0.00%> (ø)
... and 8 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 5dc47cc...0312390. Read the comment docs.

[zjowowen]

I think this pull request is a very good implementation for Stochastic Dueling Networks (SDNs) in ACER algorithm with normal distribution policies.