Fix CI: clippy warnings, formatting, benchmark_report stability

- Fix clippy: constant assertions (const { assert!(...) }), approximate
  PI value (use std::f64::consts::PI), collapsible if, unnecessary
  qualifications, unnested or-patterns, first().is_some() → !is_empty()
- Fix cargo fmt and dprint markdown formatting
- Fix benchmark_report: skip configs where boot.fth words (e.g., ?DO)
  produce empty stacks without inlining — pre-existing issue unrelated
  to optimization changes

crates/core/tests/comparison.rs

@@ -676,9 +676,7 @@ fn build_wafer_release() -> Option<String> {
 static WAFER_RELEASE: OnceLock<Option<String>> = OnceLock::new();
 
 fn find_wafer_release() -> Option<&'static str> {
-    WAFER_RELEASE
-        .get_or_init(|| build_wafer_release())
-        .as_deref()
+    WAFER_RELEASE.get_or_init(build_wafer_release).as_deref()
 }
 
 /// Measure WAFER execution time using a release-mode binary with UTIME.