fix(analyze): censor a clean run at the steps it ran, and refuse mismatched budgets

A run stops at whichever comes first, the step budget or --duration, so
a clean run that reached the wall clock exited with fewer steps than the
budget and was still credited with the whole of it. The model arm pays a
network call and a screenshot per step, so it reaches the wall sooner and
was handed exposure it never had.

Nothing checked that two arms shared a budget either. Thirty identical
clean runs under budgets of 400 and 100 read a12 0.000 and p 1.685e-14
from the rank-sum while the log-rank in the same report read p 1.0000.
groupArms already refused this within one arm.

The claims the old convention left in comments and report lines are
corrected rather than left standing beside the new behaviour.

cmd/internal-tools/analyze/ranksum_test.go

@@ -58,7 +58,7 @@ func TestVargaDelaneyA12_BoundaryCases(t *testing.T) {
 
 // The counting definition and the rank-sum route must agree, including when the
 // samples are tied against each other, which is the case the evaluation data is
-// always in because censored runs are all held at the budget.
+// usually in because censored runs pile up on the step they stopped at.
 func TestVargaDelaneyA12_AgreesWithRankSumStatistic(t *testing.T) {
 	cases := [][2][]float64{
 		{{1, 2, 3}, {2, 3, 4}},