⚡ Bolt: Optimize autograd backward pass metadata clones

[teerthsharma]

💡 What: The optimization implemented
Replaces let grad_out = grads[out.index].clone(); with Option::take() and updates accumulate_grad to take ownership of grad: Tensor.

🎯 Why: The performance problem it solves
Avoids redundant heap allocations for Tensor shape and stride metadata during the backward pass.

📊 Impact: Expected performance improvement
Eliminates one Tensor metadata clone per operation in the backprop graph, speeding up reverse-mode differentiation and reducing GC pressure.

🔬 Measurement: How to verify the improvement
Run cargo test -p aether-core to confirm autograd correctness. Performance can be measured in large-scale model training scenarios.

---

PR created automatically by Jules for task 948785078650957930 started by @teerthsharma

[google-labs-jules]

👋 Jules, reporting for duty! I'm here to lend a hand with this pull request.

When you start a review, I'll add a 👀 emoji to each comment to let you know I've read it. I'll focus on feedback directed at me and will do my best to stay out of conversations between you and other bots or reviewers to keep the noise down.

I'll push a commit with your requested changes shortly after. Please note there might be a delay between these steps, but rest assured I'm on the job!

For more direct control, you can switch me to Reactive Mode. When this mode is on, I will only act on comments where you specifically mention me with @jules. You can find this option in the Pull Request section of your global Jules UI settings. You can always switch back!

New to Jules? Learn more at jules.google/docs.

---

For security, I will only act on instructions from the user who triggered this task.