Implement GlyphOS Cognitive Kernel

Add a system service layer on top of LAIN cognition and Supercharged Glyph Registry:

Components:
- glyphos/cognitive_kernel.py: CognitiveKernel class + functional API
  * CognitiveKernel: Main orchestrator with execute_gx(), execute_symbolic()
  * Result accessors: get_last_result(), get_last_trace(), get_last_fused_symbol()
  * get_kernel(): Singleton kernel instance
  * run_gx(): Convenience function for global kernel
  * kernel_status(): Status introspection

- glyphos/__init__.py: Package initialization

- tests/test_cognitive_kernel.py: Comprehensive test suite (8 tests, 100% pass)
  * Kernel initialization and warmup
  * GX execution and result validation
  * Result accessor methods
  * Singleton pattern
  * Functional API

- COGNITIVE_KERNEL.md: Complete documentation

Test Results:
- 12 registry tests ✅
- 10 glyph bridge tests ✅
- 6 integration suites ✅
- 8 cognitive kernel tests ✅
- Total: 36 tests, 0 failures

No breaking changes - all existing tests pass.

glyphos/__init__.py

@@ -0,0 +1,19 @@
+"""GlyphOS Cognitive Kernel
+
+A system service layer on top of LAIN cognition engine and Supercharged Glyph Registry.
+Provides a clean, structured API for running cognition on GX files and managing glyph context.
+"""
+
+from .cognitive_kernel import (
+    CognitiveKernel,
+    get_kernel,
+    run_gx,
+    kernel_status,
+)
+
+__all__ = [
+    "CognitiveKernel",
+    "get_kernel",
+    "run_gx",
+    "kernel_status",
+]