Add to_dgl and from_dgl conversions

[hbenedek]

Issue #6979

Description

I add two functions from_dgl(dgl.Graph) -> data: Union[Data, HeteroData]and
to_dgl(Union[Data, HeteroData]) -> dgl.Graph in utils/convert.py. Both function in both direction performs the same three operations, setting the node and edge attributes and copies the edges of the underlying graph(s).

[codecov]

Codecov Report

Merging #7053 (7d4d9b5) into master (9ab1b05) will decrease coverage by 0.20%.
The diff coverage is 5.45%.

❗ Current head 7d4d9b5 differs from pull request most recent head 8cc3b0f. Consider uploading reports for the commit 8cc3b0f to get more accurate results

@@            Coverage Diff             @@
##           master    #7053      +/-   ##
==========================================
- Coverage   91.47%   91.27%   -0.20%     
==========================================
  Files         436      436              
  Lines       23836    23891      +55     
==========================================
+ Hits        21803    21806       +3     
- Misses       2033     2085      +52     

Impacted Files	Coverage Δ
torch_geometric/utils/convert.py	55.85% <3.70%> (-16.77%)	⬇️
torch_geometric/utils/__init__.py	100.00% <100.00%> (ø)

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[EdisonLeeeee]

Thanks for the PR! Please also includes some test cases for these two functions.

[hbenedek]

@EdisonLeeeee I added the tests, but for some reason, codecov complains about not seeing them. I ran pytest --cov and the coverage of torch_geometric/utils/convert.py is only 9%. Am I doing something wrong, or just the external libraries causing this issue?

[rusty1s]

@hbenedek Don't worry about it. It's good that we have a test but it's not called by CI due to missing dependency.

[hbenedek]

Thank you very much @rusty1s and @EdisonLeeeee, this was a lot of fun to work on :)

[rusty1s]

Great PR, thanks for doing that :)

[EdisonLeeeee]

@hbenedek thank you, too :D