Add index operators to XMFLOAT2/3/4

[ModulePillow]

Issue Description

It's valuable to add subscript operators for the XMINT2~4 and XMFLOAT2~4 types.

An example is shown below:

// For XMFLOAT4
float& operator [] (In_range(0, 3) int32_t index) noexcept { return reinterpret_cast<float*>(this)[index]; }
const float operator [] (In_range(0, 3) int32_t index) const noexcept { return reinterpret_cast<const float*>(this)[index]; }

Pros & Reasons

While I am developing my own renderer, I found that a new project relying on DirectXMath can use data types like XMFLOAT4 to store key data in its business code.

This decision has many advantages:

• Reduces memory copy. (like: a float array -> XMFLOAT4 -> XMVECTOR). The business code uses XMFLOAT4 to store key data; those operators make a vector behave like a regular array.
• Eliminates non-straightforward code. There are other weird ways to reduce memory copy, like this:

XMVECTOR v = XMLoadFloat4(reinterpret_cast<const XMFLOAT4*>(ObjA.FloatArray));

Cons

• May lack runtime boundary checks for indices.
• Abundance for some projects. For existing projects using DirectXMath as a plugin, they may not wanna use XMFLOAT4 to store core data directly.