8000 BUG: Have `norm` cast non-floating point arrays to 64-bit float arrays by jakirkham · Pull Request #7088 · numpy/numpy · GitHub