NXP backend: Use a quantized dataset for testing operators where a single-bit error is expected.

[MartinPavella]

Summary

Use a quantized dataset for testing operators where a single-bit error is expected. With a float dataset, the error depended on the output quantization. With a quantized dataset, the tolerance is always 1, regardless of quantization.

Test plan

Tested by existing nxp tests.
./backends/nxp/run_unittests.sh

cc @robert-kalmar @JakeStevens @digantdesai @rascani

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/20689

• 📄 Preview Python docs built from this PR

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

✅ No Failures

As of commit 71be8b9 with merge base 3bd4748 ():
💚 Looks good so far! There are no failures yet. 💚

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

[linux-foundation-easycla]

• ❌ - login: @MartinPavella / name: Martin Pavella. The commit (71be8b9) is not authorized under a signed CLA. Please click here to be authorized. For further assistance with EasyCLA, please visit our EasyCLA portal and chat with our support bot.

[novak-vaclav]

LGTM, good job 👌

Nit: I was just thinking that a global decorator instead of setting remove_quant_io_ops everytime might be more beautiful, but I will leave it up to you:

from functools import wraps

def allow_one_bit_error(func):
    @wraps(func)
    def wrapper(self, *args, **kwargs):
        self.remove_quant_io_ops = True
        return func(self, *args, **kwargs)
    return wrapper