Batch invariance

[jlonge4]

Issue #, if available:

N/A

Description of changes:

Demonstrates batch invariance behavior in NKI by replicating findings from Thinking Machines' "Defeating Nondeterminism in LLM Inference".

Key findings:

1. Replicated the paper: nki.lang kernels (MatMul, RMSNorm) show batch variance in both float32 and bfloat16 when using dynamic reduction strategies
2. New discovery: nki.isa operations show variance in float32 but no variance in bfloat16, making them more suitable for deterministic inference

Contents:

• MatMul and RMSNorm kernel implementations (both lang and ISA variants)
• Test suite comparing batch variance across kernel types and precisions
• Documentation explaining findings and practical implications

Testing:

Please see detailed unit test requirements in the CONTRIBUTING.md

• The change is covered by numeric check using nki.baremetal
• The change is covered by performance benchmark test using nki.benchmark
• The change is covered by end-to-end integration test

Pull Request Checklist

• I have filled in all the required field in the template
• I have tested locally that all the tests pass
• By submitting this pull request, I confirm that my contribution is made under the terms of the MIT-0 license.