feat: support deconv (1d, 2d, and Nd) dynamo converter

[zewenli98]

Description

Support deconv (1d, 2d, and Nd) dynamo converter.

Fixes #2207

Type of change

• New feature (non-breaking change which adds functionality)

Checklist:

• My code follows the style guidelines of this project (You can use the linters)
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas and hacks
• I have made corresponding changes to the documentation
• I have added tests to verify my fix or my feature
• New and existing unit tests pass locally with my changes
• I have added the relevant labels to my PR in so that relevant reviewers are notified

[gs-olive]

This looks great! Could you also add decorators for conv_transposeNd, as here, since these seem to be supported with the new implementation.

[zewenli98]

Sure! Actually I've implemented conv_transposeNd but when I did test I found they were not used while converting. For testing, do you know which PyTorch ops corresponds to these converters?

[gs-olive]

I would guess the ConvTransposeND like this operator, but these also may just end up as convolution with the boolean transpose specified. If it is not worthwhile with the argument schemas to add conv_transposeNd, then we can add a lowering pass for these

[zewenli98]

these also may just end up as convolution with the boolean transpose specified.

Yes, I found it ends up as convolution with the boolean transpose specified. That's why I didn't add conv_transposeNd. If adding conv_transposeNd, since the args are different, we need to add a new function instead of decorators. Or, just like you said, do it in lowering phase.

[gs-olive]

Looks good to me!