Add torch_geometric.nn.attention to docs

[xnuohz]

[akihironitta]

Nice! Could we have a similar list to the list of conv layers?

Convolutional Layers
--------------------

.. currentmodule:: torch_geometric.nn.conv

.. autosummary::
   :nosignatures:
   :toctree: ../generated
   :template: autosummary/nn.rst

   {% for name in torch_geometric.nn.conv.classes %}
     {{ name }}
   {% endfor %}

[xnuohz]

there are too many conv modules, if list all the details, i think searching will be inconvenient.

[akihironitta]

Wdym? nn.attention has 4 layers:

$ git grep "class " torch_geometric/nn/attention/
torch_geometric/nn/attention/performer.py:class PerformerProjection(torch.nn.Module):
torch_geometric/nn/attention/performer.py:class PerformerAttention(torch.nn.Module):
torch_geometric/nn/attention/qformer.py:class QFormer(torch.nn.Module):
torch_geometric/nn/attention/sgformer.py:class SGFormerAttention(torch.nn.Module):

[xnuohz]

misread it to expand all the conv modules haha

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 86.29%. Comparing base (3f2f43d) to head (904212b).
Report is 5 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master   #10089      +/-   ##
==========================================
- Coverage   86.74%   86.29%   -0.46%     
==========================================
  Files         493      493              
  Lines       33069    33079      +10     
==========================================
- Hits        28685    28544     -141     
- Misses       4384     4535     +151     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.