Prepare the 2.0.0 release

[smelfungus]

• toGamut with Clip no longer throws for a color ColorValue accepted. Beyond about 1e102 the conversion into LMS overflows to NaN, which clamping passed through, and the mapped color failed its own check with IllegalArgumentException. A NaN channel clamps to 0, as Css and ChromaReduction already reached by searching. A test fails on master.
• The limits of the arithmetic are stated, in KDoc and README: past about 1e102 a component maps to a color unrelated to it, and from about 1e150 to black; below an OkLCh lightness of 0.001, ChromaReduction's default solver finds the edge less precisely than the color holds, while EdgeSolver.Iterative keeps lightness and hue.
• codes.side:colorpicker stops at 1.2.0. 1.2.1 was merged but never published, and CHANGELOG and the migration table named it. Its CHANGELOG section says so.
• CHANGELOG lists 2.0's dependencies, per artifact.
• SECURITY.md gives a private reporting path, the address CODE_OF_CONDUCT.md already publishes. The Pages config drops an exclude for a folder the repository does not have.
• CI links an app against the published artifacts. tools/consumer is a build of its own that depends on colorpicker-material3 alone and links a dynamic iOS framework against what publishToMavenLocal just published, where the library's own framework is static and leaves symbol resolution to the app. With -PpickerRepo=snapshots it reads the Central snapshot repository instead.

Module	JVM	Android host	wasm
color	243	230	230
color-compose	14	—	—
colorpicker-foundation	313	—	—
colorpicker-material3	177	—	—

apiCheck and checkNoMaterial pass, and tools/consumer compiles for JVM against a local publication. The iOS link runs here in CI.