Fix the sortedness tracking for concat\_str with multiple inputs. The optimizer incorrectly marks the output of concat\_str on multiple columns as sorted, which can cause wrong results in downstream operations that rely on sorted invariants. The fix should not propagate sortedness metadata when concat_str operates on multiple input columns, since concatenating multiple sorted columns does not produce a sorted result.

Implement the fix. Run the test suite with the test command to verify.

Fix inconsistent behavior when dividing literals by zero. Currently, pl.lit(1) // pl.lit(0) produces different results depending on the dtype: panics for unsigned types, returns 0 for same-type signed ints, and returns null for mixed types. The correct behavior is to always return null when dividing by zero, matching standard SQL/dataframe semantics. Fix the literal simplification logic in the optimizer to handle division-by-zero cases consistently.

Implement the fix. Run the test suite with the test command to verify.