Add documentation for Hadamard product with different memory layouts (PR #37)

[Copilot]

Overview

This PR adds comprehensive documentation for the Hadamard product feature implemented in PR #37, which enables element-wise matrix multiplication for matrices with different memory layouts.

What's Documented

Added a new "Hadamard Product (Element-wise Multiplication)" section to site/docs/matrix.md that covers:

Basic Usage

Shows simple element-wise multiplication of two matrices:

val mat1 = Matrix.fromRows(
  NArray[Double](1.0, 2.0, 3.0),
  NArray[Double](4.0, 5.0, 6.0)
)
val mat2 = Matrix.fromRows(
  NArray[Double](2.0, 3.0, 4.0),
  NArray[Double](5.0, 6.0, 7.0)
)
val result = mat1.hadamard(mat2)

Advanced Scenarios

Non-contiguous Views: Demonstrates Hadamard product on sliced matrix views, which previously would throw NotImplementedError:

val base = Matrix[Double](NArray.tabulate[Double](9)(i => (i + 1).toDouble), 3, 3)
val view = base(::, NArray(1, 2))  // Select specific columns
val result = view.hadamard(view)   // Now works!

Mixed Layouts: Shows operations between matrices with different layouts (one simple, one sliced)

Transposed Matrices: Demonstrates Hadamard product with row-major (transposed) matrices

Performance Insights

Documents the optimization strategies:

• Fast path: SIMD-optimized array multiplication when both matrices have the same dense layout
• Different layouts: Intelligent materialization of only one matrix to match the other's layout
• Cross-platform: SIMD on JVM (using Java's Vector API), efficient while loops on JS and Native

Changes Made

• Added 119 lines of documentation to site/docs/matrix.md
• All code examples are executable and verified through mdoc
• Documentation builds successfully with ./millw site.siteGen
• All tests pass on JVM platform

Why This Matters

PR #37 removed a significant limitation where the Hadamard product would fail with NotImplementedError for non-simple matrix layouts. This documentation helps users understand:

• That they can now freely use Hadamard product with sliced views, transposed matrices, and mixed layouts
• How vecxt intelligently handles the complexity behind the scenes
• The performance characteristics of different scenarios

This aligns with vecxt's goal of providing "reasonable, consistent cross platform ergonomics" while maintaining correctness as the #1 priority.

Original prompt

---

This section details on the original issue you should resolve

<issue_title>Write documentation for #37</issue_title>
<issue_description></issue_description>

Comments on the Issue (you are @copilot in this section)

Fixes #38

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.