[WIP] Support PyTorch-lightning DDP training

[tohmae]

This PR addresses issue #2699.
This is related to Lightning-AI/pytorch-lightning#7916.
So DDP training on multi GPUs supports from pytorch-lightning version 1.4.0.

[codecov-commenter]

Codecov Report

Merging #2824 (0a48595) into master (e0fd830) will decrease coverage by 0.31%.
The diff coverage is 87.50%.

@@            Coverage Diff             @@
##           master    #2824      +/-   ##
==========================================
- Coverage   91.79%   91.48%   -0.32%     
==========================================
  Files         146      147       +1     
  Lines       11192    11368     +176     
==========================================
+ Hits        10274    10400     +126     
- Misses        918      968      +50     

Impacted Files	Coverage Δ
optuna/integration/pytorch_lightning.py	88.23% <87.50%> (-0.66%)	⬇️
optuna/cli.py	26.97% <0.00%> (-2.60%)	⬇️
optuna/study/_optimize.py	97.77% <0.00%> (-0.04%)	⬇️
optuna/storages/__init__.py	100.00% <0.00%> (ø)
optuna/storages/_heartbeat.py	100.00% <0.00%> (ø)
optuna/samplers/_tpe/sampler.py	96.65% <0.00%> (+0.01%)	⬆️
optuna/samplers/_tpe/parzen_estimator.py	98.93% <0.00%> (+0.03%)	⬆️
optuna/storages/_cached_storage.py	98.69% <0.00%> (+1.62%)	⬆️
optuna/visualization/matplotlib/_contour.py	83.21% <0.00%> (+3.53%)	⬆️

---

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 e0fd830...0a48595. Read the comment docs.

[tohmae]

I am sorry. This PR is wrong so I closed it.