Review follow-up: changes required on PR #84

[BnJam]

Code review requested changes on PR #84 before merge. Required changes:

• Major: spike/outlier split points are silently dropped (IEEE-754 simulation: spike series yields segments [0,24] and [26,49], index 25 missing), contradicting the PR's validation claim.
• Major: recursion depth is still O(n) worst-case (n=10000 convex series -> depth ~4052), so large inputs can still overflow the stack.
• Minor: 2-element input with threshold=0.0 returns zero segments.
• Minor: no new regression tests; existing test only checks len >= 2.