Optimize shape rendering (pen creation mostly).

[MarchingCube]

What does the pull request do?

An optimization for core Shape rendering that minimizes perf and memory impact.
Best case (no brushes) improved cpu perf x200 and removed all allocations.
Worst case we are looking at cpu perf x38 and 98% memory reduction.

What is the current behavior?

Method	Mean	Error	StdDev	Gen 0	Gen 1	Gen 2	Allocated
Render_Line_NoBrushes	9.246 us	0.0540 us	0.0506 us	0.7019	0.0153	-	3.01 KB
Render_Line_WithStroke	14.182 us	0.0617 us	0.0577 us	0.7782	0.0305	-	3.32 KB
Render_Line_WithFill	9.320 us	0.0447 us	0.0373 us	0.7019	0.0153	-	3.01 KB
Render_Line_WithFillAndStroke	13.953 us	0.0374 us	0.0312 us	0.7782	0.0305	-	3.32 KB

What is the updated/expected behavior with this PR?

Method	Mean	Error	StdDev	Gen 0	Gen 1	Gen 2	Allocated
Render_Line_NoBrushes	45.30 ns	0.092 ns	0.086 ns	-	-	-	-
Render_Line_WithStroke	339.26 ns	1.937 ns	1.812 ns	0.0210	-	-	88 B
Render_Line_WithFill	87.35 ns	0.374 ns	0.332 ns	-	-	-	-
Render_Line_WithFillAndStroke	352.55 ns	2.207 ns	2.065 ns	0.0210	-	-	88 B

How was the solution implemented (if it's not obvious)?

Best optimization is to just avoid doing any work, so when Stroke is not set we can just ignore it.
Also majority of rendered shapes are not using dashes so there is no point in allocating dash style for everything.

Biggest change is using immutable pen and dash styles. Mutable variants have an absurd overhead due to change tracking.

Checklist

• Added unit tests (if possible)?
• Added XML documentation to any related classes?
• Consider submitting a PR to https://github.com/AvaloniaUI/Documentation with user documentation

Breaking changes

Obsoletions / Deprecations

Fixed issues