Fix for Type Check in Quantized CPU op_dequantize

[BujSet]

Summary

When quantizing a model (without delegating to a specific backend), an exported model relies on the operator library in kernels/quantized/cpu/. Specifically, the essential operation of op_dequantize is performing:

out = (in - offset) * scale

where the offset is an integer type. While initially, this offset is assumed to be an uint64_t (see here), when it is used to perform the operation above, it is cast down to a uint32_t (see here). It seems an implicit assumption is that the quantization offset is a uint32_t value, and the uint64_t declaration is simply safeguarding for future proofing. In any event, the type check for the offset should allow the offset to be either uint32_t or uint64_t`. This PR allows for that change.

Test plan

Tested with mobilenet V2 on Arm backend. Quantized model runner initially crashed do to this check only allowing the offset to be uint64_t. When examining the values, none were larger than UINT32_MAX, so it should be safe to permit the offset to have uint32_t values. When this change was made, the mobilenet V2 runner was able to complete.

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/13174

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 2 New Failures, 1 Unrelated Failure

As of commit bc3d730 with merge base be221c6 ():

NEW FAILURES - The following jobs have failed:

• Build documentation / build (buck2) / Build doc (gh)
  At least one of the pre-conditions you specified did not hold
• pull / test-moshi-linux / linux-job (gh)
  RuntimeError: Command docker exec -t f0e0d85a106c616eced5d4641f6ff69eb97648eb3ce2dfa47135f006e7be0d49 /exec failed with exit code 1

BROKEN TRUNK - The following job failed but were present on the merge base:

👉 Rebase onto the `viable/strict` branch to avoid these failures

• pull / test-openvino-linux / linux-job (gh) (trunk failure)
  AttributeError: '_OpNamespace' 'quantized_decomposed' object has no attribute 'convert_element_type'

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[BujSet]

@pytorchbot label "release notes: none"