update torchsharp to 0.105.0

[LittleLittleCloud]

We are excited to review your PR.

So we can do the best job, please check:

• There's a descriptive title that will make sense to other developers some time from now.
• There's associated issues. All PR's should have issue(s) associated - unless a trivial self-evident change such as fixing a typo. You can use the format Fixes #nnnn in your description to cause GitHub to automatically close the issue(s) when your PR is merged.
• Your change description explains what the change does, why you chose your approach, and anything else that reviewers should know.
• You have included any necessary tests in the same PR.

[Copilot]

Pull Request Overview

This PR updates the TorchSharp integration to version 0.105.0 by refining API calls and enforcing non-null assertions on buffers. Key changes include:

• Adding null-forgiving operators ("!") to get_buffer calls in both RotaryEmbedding.cs and QuantizedLinear.cs.
• Updating named parameter "kernelSize" to "kernel_size" in ConvModule.cs to align with TorchSharp’s API.

Reviewed Changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated no comments.

File	Description
src/Microsoft.ML.GenAI.Core/Module/RotaryEmbedding.cs	Added null-forgiving operator on get_buffer usage.
src/Microsoft.ML.GenAI.Core/Module/QuantizedLinear.cs	Added null-forgiving operator on multiple get_buffer calls and adjusted arithmetic operations.
src/Microsoft.ML.TorchSharp/AutoFormerV2/ConvModule.cs	Renamed named parameter for consistency with TorchSharp.

[ericstj]

Which feed are these new versions published to? If only nuget.org, should they be mirrored?

[michaelgsharp]

Yes it needs mirroring. Being done here, https://dev.azure.com/dnceng/internal/_build/results?buildId=2676108&view=results.

[michaelgsharp]

/azp run

[azure-pipelines]

Azure Pipelines successfully started running 2 pipeline(s).

[michaelgsharp]

Ah, I think I remember now. The reason we had stayed on the old version was that these newer versions don't support OSX-x64. So its gonna require a bit more work to swap versions @LittleLittleCloud. Going to have to stop torch sharp from building on osx x64 and skip those tests there as well. We've dont something similar in ML.NET for x86, but now it will need to be expanded to this as well.