fix(analyze): divide by actions that ran

Defects per thousand actions counted every step, including steps that chose
nothing and steps whose action was never dispatched. The inflation is
policy-dependent, so it does not cancel between arms: on the fixture campaign
the model arm's yield was reported at 60.3 per thousand against a true 120.7,
because half its steps did nothing.

A runs.jsonl without the count is refused by name and line rather than read as
zero actions, which would report every per-action rate wrongly. The report also
carries steps beside actions now, so the gap is visible rather than folded into
a denominator.

Claude-Session: https://claude.ai/code/session_01A5KmftdEJ49A9z5mF5ESrX

cmd/internal-tools/analyze/report.go

@@ -32,11 +32,13 @@ func writeReport(result analysis, out io.Writer) {
 
 	fmt.Fprintln(out)
 	fmt.Fprintln(out, "a detection is one distinct property violated in one run; run hours sum the per-run wall clock")
-	writeTable(out, []string{"arm", "actions", "run hours", "detections", "defects/1k actions", "defects/hour", "distinct defects", "found in one run"},
+	fmt.Fprintln(out, "actions count the steps that dispatched one; the rest chose nothing or had the choice thrown away")
+	writeTable(out, []string{"arm", "steps", "actions", "run hours", "detections", "defects/1k actions", "defects/hour", "distinct defects", "found in one run"},
 		func(add func(...string)) {
 			for _, summary := range result.Arms {
 				add(
 					summary.Arm,
+					strconv.Itoa(summary.TotalSteps),
 					strconv.Itoa(summary.TotalActions),
 					fmt.Sprintf("%.2f", summary.TotalRunHours),
 					strconv.Itoa(summary.Detections),