ok    1  best-fit line: slope 2, intercept 10 (exact)
ok    2  x = lstsq(...) leaves _integer alone (2026-09-19)
ok    3  lstsq == solve(matmul(transpose(a), a), matmul(transpose(a), b))
ok    4  square full-rank reduces to solve: 2.5, 2.75 (exact)
ok    5  square: lstsq(a, b) = solve(a, b)
ok    6  non-terminating fit: slope 1.5 exact
ok    7  non-terminating intercept 1/3 within 1e-15
ok    8  m x k rhs: two fits at once (second is double the first)
ok    9  stats$sum(lstsq(...)) = 2 + 10
ok    10  arithmetic on lstsq(): * 10
ok    11  lstsq inside an f$ slot
ok    12  month 7 forecast from the fit: 24
ok    13  rank-deficient a raises SINGULAR
ok    14  b with the wrong row count raises ARRAYSHAPE
ok    15  scalar-context lstsq raises ARRAYRESULT

passed:  15   failed:  0 
