Fix Conv LHS packing padding/uninitialized ptrs V2

[hariharans29]

Description

Refer to V1 of the fix here: #27214

This PR includes all fixes from the V1 PR + logic to invalidate the lhs cache pointers in case the pad buffer's underlying buffer has changed due to a resize. The ARM team will look at potentially enhancing this logic after the 1.24.0 release.

Motivation and Context

Fix #26669

[Copilot]

Pull request overview

Addresses non-deterministic correctness issues by hardening KleidiAI Conv LHS packing/padding behavior and adding an additional CUDA ConvTranspose bias validation.

Changes:

• Initialize all entries in the KleidiAI Conv LHS indirection table to padding pointers to avoid uninitialized reads for partial tiles.
• Replace the shared static padding buffer with a thread_local buffer and invalidate cached LHS pointer tables when the padding buffer reallocates.
• Add CUDA ConvTranspose bias shape validation against the computed output channel count.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 3 comments.

File	Description
onnxruntime/core/providers/cuda/nn/conv_transpose.cc	Adds runtime validation that bias is a 1-D tensor matching num_output_channels.
onnxruntime/core/mlas/lib/kleidiai/convolve_kleidiai.cpp	Initializes LHS pointer table padding entries and updates padding-buffer + cache invalidation behavior.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated 1 comment.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.