Refocus on wall-clock time and add cached clean benchmark for COMPILATION_CACHING

[AvdLee]

Summary

• Refocus all skills on wall-clock build time as the primary success metric. Cumulative task-time savings are diagnostic evidence, not proof of wait-time improvement. Every recommendation now states its expected wall-clock impact using standardized impact language.
• Add cached clean benchmark that measures clean builds with a warm compilation cache. Standard clean builds show cache-population overhead without cache-hit benefit, making COMPILATION_CACHING appear to regress performance. The new phase deletes DerivedData (not the system compilation cache) between runs, capturing the realistic developer workflow: branch switching, pulling changes, and Clean Build Folder.

Changes across 15 files

Scripts:

• benchmark_builds.py: auto-detects COMPILATION_CACHING = YES and runs cached clean builds using a controlled DerivedData path. New --no-cached-clean flag to opt out.
• generate_optimization_report.py: shows "Cached Clean" column in baseline table when data is present; adds timing summary section for cached clean builds.

Schema:

• build-benchmark.schema.json: bumped to 1.2.0 with optional cached_clean in runs/summary and "cached-clean" build type.

Skills & references:

• All six skill SKILL.md files updated with wall-clock-first prioritization and cached clean awareness.
• recommendation-format.md: added required wait_time_impact field.
• benchmark-artifacts.md: replaced "trust us" COMPILATION_CACHING caveat with cached clean documentation.
• build-settings-best-practices.md: added measurement note to COMPILATION_CACHING entry.
• orchestration-report-template.md: expanded baseline table, verification section, and community row template for cached clean.
• README.md: community results table includes Cached Clean columns.

Test plan

• Run benchmark_builds.py on a project without COMPILATION_CACHING and verify output matches schema 1.1.0 (no cached_clean key)
• Run benchmark_builds.py on a project with COMPILATION_CACHING = YES and verify cached clean runs appear in the artifact
• Run generate_optimization_report.py with both artifact types and verify the Cached Clean column appears only when data exists
• Verify --no-cached-clean flag skips the cached clean phase even when COMPILATION_CACHING is detected