Stream: reduce & Aggregate

Terminal operations trigger the lazy pipeline and consume the stream: collect accumulates, reduce folds, forEach acts, count counts, findFirst returns the first match, and anyMatch checks a condition.

forEach has no order guarantee in parallel (fast); forEachOrdered preserves encounter order in parallel (slower); both are ordered in sequential streams.

findFirst returns the deterministic first element in order (parallel: coordination cost); findAny returns any match non-deterministically (parallel: fastest thread wins); both are identical in sequential.

With an identity you always get a real value; without an identity you get an Optional because "no elements" means "no answer".

LongStream.rangeClosed(1, n).reduce(1, Math::multiplyExact) computes factorial with overflow detection, throwing ArithmeticException instead of wrapping silently.

`reduce(ZERO, (sum, e) -> sum.add(e.salary), BigDecimal::add)` folds every salary into one grand total.

`reduce((a, b) -> a.salary >= b.salary ? a : b)` keeps the higher earner at each step and returns an Optional.

reduce from an empty string and append each name, inserting a comma only when the running string is not empty.

`reduce(1L, (a, b) -> a * b)` multiplies all ages, and the identity must be 1, never 0.