[FEATURE SUPPORT] add triton geru kernel

[LoserCheems]

Summary

• Introduces a Triton kernel for in-place outer-product updates on matrices, enhancing performance for alpha-scaled vector outer products.

Root Cause

• The previous implementation lacked an efficient method for updating strided tiles, leading to performance bottlenecks.

Changes

• Added a new Triton kernel for the GERU operation and updated the README to reflect the completion of the Triton implementation.

Reproduction

• Not applicable as this is a new feature addition.

Tests

• Existing tests were updated to include the new Triton kernel implementation.

Compatibility

• No migration concerns or backwards compatibility issues identified.

Checklist

• Linked issue provided [FEATURE REQUEST] geru Triton kernel implementation #27
• Adds or updates tests
• Updates docs if needed
• No perf regressions