@register_* decorator in GraphGym

[rusty1s]

Adds decorator functionality to register GraphGym modules:

@register_layer('linear')
class Linear(torch.nn.Module):
    pass

[codecov-commenter]

Codecov Report

Merging #3684 (351c2fc) into master (cf6bedf) will increase coverage by 0.11%.
The diff coverage is 94.68%.

@@            Coverage Diff             @@
##           master    #3684      +/-   ##
==========================================
+ Coverage   81.57%   81.68%   +0.11%     
==========================================
  Files         295      295              
  Lines       14880    14893      +13     
==========================================
+ Hits        12138    12165      +27     
+ Misses       2742     2728      -14     

Impacted Files	Coverage Δ
torch_geometric/graphgym/register.py	90.38% <90.38%> (+19.27%)	⬆️
...orch_geometric/graphgym/contrib/train/benchmark.py	21.31% <100.00%> (ø)
torch_geometric/graphgym/model_builder.py	92.85% <100.00%> (-0.48%)	⬇️
torch_geometric/graphgym/models/act.py	100.00% <100.00%> (ø)
torch_geometric/graphgym/models/encoder.py	31.81% <100.00%> (-1.52%)	⬇️
torch_geometric/graphgym/models/gnn.py	80.26% <100.00%> (ø)
torch_geometric/graphgym/models/head.py	50.00% <100.00%> (+1.78%)	⬆️
torch_geometric/graphgym/models/layer.py	73.11% <100.00%> (+1.36%)	⬆️
torch_geometric/graphgym/models/pooling.py	100.00% <100.00%> (+46.15%)	⬆️

---

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 cf6bedf...351c2fc. Read the comment docs.

[JiaxuanYou]

This makes the implementation a lot cleaner. Thanks!