Add NodeFormer model and example

[xnuohz]

Issue

• From Graph Transformer Enhancement #9751
• Adds an example for training NodeFormer model on ogbn-arxiv dataset.

Usage

python examples/ogbn_train.py --model nodeformer

Highlight

Result

• GAT and SAGE use batch_size=1024
• SGFormer and Polynormer use batch_size=512, otherwise OOM

Dataset	GAT	SAGE	SGFormer	Polynormer	NodeFormer
ogbn-arxiv	68.53%	70.17%	71.86%	65.90%	66.98%

[codecov]

Codecov Report

❌ Patch coverage is 81.11888% with 54 lines in your changes missing coverage. Please review.
✅ Project coverage is 84.16%. Comparing base (c211214) to head (272b7ea).
⚠️ Report is 191 commits behind head on master.

Files with missing lines	Patch %	Lines
torch_geometric/nn/models/nodeformer.py	81.05%	54 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master   #10409      +/-   ##
==========================================
- Coverage   86.11%   84.16%   -1.96%     
==========================================
  Files         496      511      +15     
  Lines       33655    36305    +2650     
==========================================
+ Hits        28981    30555    +1574     
- Misses       4674     5750    +1076     

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[puririshi98]

@xnuohz plz get CI passing

[puririshi98]

LGTM just get CI green

[puririshi98]

@rusty1s @akihironitta @wsad1 ready for final review/merge

[puririshi98]

@xnuohz can you explain where we are ensuring attention is not being done across the full minibatch? we should be utilizing the batch tensor, no? (remember what we had to change for sgformer)