bench: capture T0 baseline; expose pi/tau accessors; fix div panic

- Promotes Gaussian::pi and Gaussian::tau to public so benches/gaussian.rs
  compiles, then captures baseline numbers for the T0 acceptance gate.
- Fixes the divide bench: g1/g2 panicked (g1 has lower precision than g2;
  cavity requires pi_num >= pi_den). Swapped to g2/g1 (well-defined).

Baseline on Apple M5 Pro:
  Batch::iteration  29.840 µs
  Gaussian::mul      1.568 ns   (vs ~220 ps for add/sub — hot path)
  Gaussian::div      1.572 ns

benches/baseline.txt

@@ -0,0 +1,12 @@
+# Baseline numbers captured before T0 changes
+# Hardware: lrrr.local / Apple M5 Pro
+# Date: 2026-04-24
+
+Batch::iteration          29.840 µs
+Gaussian::add            219.58 ps
+Gaussian::sub            219.41 ps
+Gaussian::mul              1.568 ns
+Gaussian::div              1.572 ns
+Gaussian::pi             262.89 ps
+Gaussian::tau            262.47 ps
+Gaussian::pi_tau_combined 219.40 ps

benches/gaussian.rs

@@ -23,8 +23,11 @@ fn benchmark_gaussian_arithmetic(criterion: &mut Criterion) {
     });
 
     // Benchmark division
+    // NOTE: numerator must have higher precision (smaller sigma) than the
+    // denominator in this representation; g2 (sigma=1) / g1 (sigma=8.33) is
+    // well-defined, whereas g1 / g2 underflows and panics in mu_sigma.
     criterion.bench_function("Gaussian::div", |bencher| {
-        bencher.iter(|| g1 / g2);
+        bencher.iter(|| g2 / g1);
     });
 
     // Benchmark natural parameter conversions

src/gaussian.rs

@@ -13,7 +13,7 @@ impl Gaussian {
         Gaussian { mu, sigma }
     }
 
-    fn pi(&self) -> f64 {
+    pub fn pi(&self) -> f64 {
         if self.sigma > 0.0 {
             self.sigma.powi(-2)
         } else {
@@ -21,7 +21,7 @@ impl Gaussian {
         }
     }
 
-    fn tau(&self) -> f64 {
+    pub fn tau(&self) -> f64 {
         if self.sigma > 0.0 {
             self.mu * self.pi()
         } else {