Bug Fix: Order of transforms in label_prop example

[Riyer01]

This example doesn't work properly as currently written. The correct way is to reverse the order of transformations, such that we first convert the graph to undirected, then get the sparse representation.

Empirically, try running this file with the transformations listed both ways for 50 iterations, and it becomes clear from the results that the original implementation is incorrect.

Submitted for extra credit in CS224W

[codecov-commenter]

Codecov Report

Merging #3639 (2ac3644) into master (e8915ad) will decrease coverage by 0.18%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #3639      +/-   ##
==========================================
- Coverage   81.66%   81.47%   -0.19%     
==========================================
  Files         294      294              
  Lines       14655    14745      +90     
==========================================
+ Hits        11968    12014      +46     
- Misses       2687     2731      +44     

Impacted Files	Coverage Δ
torch_geometric/loader/dataloader.py	80.55% <0.00%> (-2.78%)	⬇️
torch_geometric/loader/utils.py	86.76% <0.00%> (-2.13%)	⬇️
torch_geometric/typing.py	100.00% <0.00%> (ø)
torch_geometric/data/__init__.py	100.00% <0.00%> (ø)
torch_geometric/nn/models/basic_gnn.py	89.36% <0.00%> (ø)
torch_geometric/data/lightning_data_module.py
torch_geometric/data/lightning_datamodule.py	25.66% <0.00%> (ø)
torch_geometric/transforms/gdc.py	78.97% <0.00%> (+0.10%)	⬆️
torch_geometric/transforms/to_undirected.py	97.43% <0.00%> (+0.13%)	⬆️
torch_geometric/loader/hgt_loader.py	92.50% <0.00%> (+2.02%)	⬆️
... and 1 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 e8915ad...2ac3644. Read the comment docs.

[rusty1s]

You are right. Thanks for spotting!