Fix hang in constant folding when Resize output exceeds size limit

[Copilot]

Constant folding hangs on Resize nodes (and potentially other ops) where all inputs are constants but the output is very large — e.g., Resize with mode=cubic upscaling 1×96×7×7 → 1×96×256×256. The ONNX reference implementation for cubic interpolation is extremely slow at this scale, and the output_size_limit guard only fired after evaluation completed (in _prepare_folded_tensor), so the evaluator was never interrupted.

Changes

• _constant_folding.py: In FoldConstantsPass.process_node, added a pre-evaluation output size check. Before invoking _reference_evaluator.evaluate, the code now estimates the output tensor size from statically-known output shapes. If any output exceeds output_size_limit, the node is skipped immediately — no evaluation is attempted.

• _constant_folding_test.py: Added regression test test_output_size_limit_prevents_evaluation_before_running — creates a Resize node with all-constant inputs and an output that exceeds a small output_size_limit, verifying the node is not folded (and does not hang).

Original prompt

---

This section details on the original issue you should resolve

<issue_title>Infinite loop in constant folding of Resize layer</issue_title>
<issue_description>converting the attached ONNX model

with the following script:

from onnxscript import ir
from onnxscript.optimizer import optimize

ir_model = ir.load("input.onnx")
optimized_model = optimize(ir_model)
ir.save(optimized_model, "output.onnx")

results in a hang when constant folding the Resize layer

input.onnx.zip

versions of relevant packages:

ml_dtypes         0.5.4
onnx              1.20.1
onnx-ir           0.2.0
onnxscript        0.6.2
```</issue_description>

## Comments on the Issue (you are @copilot in this section)

<comments>
</comments>

• Fixes Infinite loop in constant folding of Resize layer #2859

---

🔒 GitHub Advanced Security automatically protects Copilot coding agent pull requests. You can protect all pull requests by enabling Advanced Security for your repositories. Learn more about Advanced Security.

[codecov]

Codecov Report

❌ Patch coverage is 87.50000% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 71.85%. Comparing base (4c4f7a0) to head (fdaa406).
⚠️ Report is 2 commits behind head on main.
✅ All tests successful. No failed tests found.

Files with missing lines	Patch %	Lines
onnxscript/optimizer/_constant_folding.py	80.00%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2860      +/-   ##
==========================================
- Coverage   71.86%   71.85%   -0.02%     
==========================================
  Files         239      239              
  Lines       29139    29154      +15     
  Branches     2875     2878       +3     
==========================================
+ Hits        20942    20949       +7     
- Misses       7219     7229      +10     
+ Partials      978      976       -2     

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