Only claim conv overloads the QAT fold can actually match

[billmguo]

Summary:
annotate_conv_bn_partitions listed aten.conv2d.padding and
aten.convolution.default in CONV_TARGETS, but torchao can never fold either.
_fuse_conv_bn_qat and _fold_conv_bn_qat both trace their patterns from
F.conv1d, F.conv2d, F.conv_transpose1d and F.conv_transpose2d, so only
those four overloads can match.

Claiming a conv the fold cannot match is worse than not claiming it. The pass
drops the conv's output qspec on the assumption the fold will move the bn output
observer onto the conv; when the fold misses, the conv is left emitting float32
into a quantized graph and QNN rejects it with 0x232 != 0x408 (FLOAT_32 meeting
UFIXED_POINT_8) -- the very failure the module docstring describes.

nn.Conv2d(..., padding="same") exports to aten.conv2d.padding and hit exactly
this. aten.convolution.default only appears after decomposition, long after
this pass runs, so it never matched anything.

Measured on master, conv+bn+relu through prepare_qat_pt2e -> convert_pt2e
with QnnQuantizer(8a8w, is_qat=True):

model	exported target	in CONV_TARGETS	bn survives	conv output unquantized
Conv2d(padding=1)	conv2d.default	yes	no	0
Conv2d(padding="same")	conv2d.padding	yes	yes	1
Conv1d(padding="same")	conv1d.padding	no	yes	0
ConvTranspose2d	conv_transpose2d.input	yes	no	0

conv1d.padding is the control: it also fails to fold -- that is a torchao
limitation this pass cannot fix -- but because it was never claimed, the per-node
annotator still gives the conv an output qspec and the graph stays valid. Only
the overload this pass claimed lost its output. After narrowing CONV_TARGETS,
the conv2d.padding case reports 0 unquantized conv outputs.

Note _is_conv_or_conv_transpose_node is not a usable guard here -- it returns
True for conv2d.padding (torchao/quantization/pt2e/utils.py, _is_conv_node).

Differential Revision: D117778197

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/22241

• 📄 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, 7 Pending, 1 Unrelated Failure

As of commit d4e9d75 with merge base d0c2da3 ():

NEW FAILURES - The following jobs have failed:

• Cadence Build & Test / hifi-build / hifi4 (gh)
  ##[error]Refusing to check out fork pull request code from a 'pull_request_target' workflow. This workflow runs with the base repository's GITHUB_TOKEN, secrets, default-branch cache scope, and runner access. Fetching and executing a fork's code in that trusted context commonly leads to "pwn request" vulnerabilities. To opt in, review the risks at https://gh.io/securely-using-pull_request_target and set 'allow-unsafe-pr-checkout: true' on the actions/checkout step.
• Cadence Build & Test / vision-build / vision (gh)
  ##[error]Refusing to check out fork pull request code from a 'pull_request_target' workflow. This workflow runs with the base repository's GITHUB_TOKEN, secrets, default-branch cache scope, and runner access. Fetching and executing a fork's code in that trusted context commonly leads to "pwn request" vulnerabilities. To opt in, review the risks at https://gh.io/securely-using-pull_request_target and set 'allow-unsafe-pr-checkout: true' on the actions/checkout step.

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-qnn-testsuite-linux / test-backend-linux (qnn, models) / linux-job (gh) (trunk failure)

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

[meta-codesync]

@billmguo has exported this pull request. If you are a Meta employee, you can view the originating Diff in D117778197.

[github-actions]

This PR needs a release notes: label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

[qti-horodnic]

LGTM, thanks for fixing the issue quickly