[Profiling] Fix performance regression in FastTextRenderer

[Eideren]

PR Details

@manio143 found a performance regression introduced in #1896 - this PR fixes it.

Description

Mapped resources are typically uncached with write combining - so reading from the buffer is significantly slower.
Issue came up only when a debugger with a disable JIT on load flag is hooked to the process.
The pre JIT'ed assembly for GraphicsFastTextRendererGenerateVertices had unnecessary reads built in, the attached debugger prevented further optimization, leading to a bunch of extremely slow reads on the mapped resource.

This PR also:

• Strips out Position.Z and Normals from the vertex buffer used to improve throughput.
• Uses Numerics' vector for SIMD to build the vertex buffer.
• Removes an unnecessary memory clear since everything that's shown on screen is assigned through the vertex buffer loop.
• Removes the unintended public access modifier for GraphicsFastTextRendererGenerateVertices, it was internal before PR [Native] - Implement some existing C++ methods in C# #1896.
• Move a couple of fields from class to function.
• Introduce a new structure, Vertex2DPositionTexture

Related Issue

Supersedes PR #2049

Types of changes

• Docs change / refactoring / dependency upgrade
• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)

Checklist

• My change requires a change to the documentation.
• I have added tests to cover my changes.
• All new and existing tests passed.
• I have built and run the editor to try this change out.