Expected range of a random walk in Lean: count last visits, skip time reversal

To prove E #{S_0..S_n}/(n+1) -> P(escape) for an iid walk: (1) deterministic: #((range (n+1)).image s) = #{k ∈ range (n+1) | ∀ j ∈ Ioc k n, s j ≠ s k} via card_nbij s; surjectivity picks the last index with Finset.max' of {j | s j = s i}. (2) The event 'k is a last visit' equals {(ξ_{k+t})_t ∈ avoidUpTo (n-k)} (sum_range_add), so by shift invariance its probability is a_{n-k}; sum_range_reflect turns the sum into a_0 + ... + a_n. (3) a_m decreases to P(escape) by tendsto_measure_iInter_atTop (avoidUpTo is antitone, ⋂ = escape), then ENNReal.tendsto_toReal. (4) Filter.Tendsto.cesaro composed with tendsto_add_atTop_nat 1 finishes. Whole file ~150 lines; textbook time reversal is unnecessary.