[LINALG] Fix: Incorrect linalg lowering for aten.convolution_transpose with negative effective padding

[hariprasadravi]

The Bug

The torch-to-linalg lowering for aten.convolution (with transposed=true) incorrectly handles cases where the effective padding is negative. The logic for this is contained in createTransposedInputPadding.

The original implementation had two critical flaws:

Incorrect Math: The logic block for negative padding (if (anyDimensionPaddingIsNegative)) attempted to "pre-crop" the input tensor before un-striding. The math used to calculate these slice offsets and sizes was incorrect, resulting in tensor.extract_slice operations with out-of-bounds offsets and negative sizes, causing the compiler to fail.

Failed "Mixed-Mode" Logic: The code was built on an "all-or-nothing" assumption. It failed to handle "mixed-mode" padding, where one spatial dimension required padding (positive offset) while another required cropping (negative offset). It would enter the negative padding path and apply cropping logic to all dimensions, leading to out-of-bounds errors when it tried to crop a dimension that should have been padded.

The Fix

This patch refactors the logic into two clean, robust paths:

All-Padding Path (else block):

Trigger: All spatial dimensions have an effective padding offset >= 0.

Action: Retains the original, efficient "fast path." It uses a single tensor.insert_slice to perform both un-striding (with strides) and padding (with offsets) in one operation.

Safe Path (if (anyDimensionPaddingIsNegative) block):

Trigger: At least one spatial dimension has a negative effective padding offset.

Action: This path is now a unified, robust 3-step process that correctly handles both all-crop and mixed-mode scenarios:

Create "Super-Tensor": It computes a maxSizes tensor, which is the "union" of the padded and un-strided sizes (i.e., max(innerSize, outerSize) for each dimension).

Pad & Un-stride: It performs a single tensor.insert_slice of the original input into this maxSizes tensor. This one operation correctly applies all positive padding (via insertSliceOffsets) and un-striding (via strideIndexValues).

Crop: It performs a final tensor.extract_slice to crop the maxSizes tensor down to the final outerSizes. This correctly applies all negative padding (via extractSliceOffsets).

This new logic resolves all known failure cases and is validated by the new TransposedConv{1,2,3}dNegativePadding test cases, which specifically target this functionality.

[hariprasadravi]

@sahas3 and @ivangarcia44 Please help review and add additional reviewers.

[ivangarcia44]

In general LGTM. Just minor comments.

[ivangarcia44]

The changes LGTM. No more comments on my side. Thank you for adding the description. It is easy to see now in the formula why a large stride is going to make a negative padding.

[zjgarvey]

Thanks for the very clear PR. I don't have any comments, and the work looks correct to me.

[sahas3]

LG! Thanks for fixing this.