Infection spread (multi-part simulation)

A rectangular grid holds 0 (healthy) and 1 (infected). Each day, every infected cell infects its healthy neighbours up, down, left and right (no diagonals). All cells update at the same time: a cell infected today only starts spreading tomorrow.

Implement time_to_full_infection(grid) -> int:

• Return the number of days until no healthy cell is left.
• Return 0 if there is no healthy cell to begin with (this includes an empty grid, [] or [[]]).
• Return -1 if some healthy cell can never be infected (here that only happens when there is no infected cell at all).
• Don't modify grid.

time_to_full_infection([[0, 0, 0],
                        [0, 1, 0],
                        [0, 0, 0]])     # 2: edges on day 1, corners on day 2
time_to_full_infection([[0, 0, 0, 1]])  # 3
time_to_full_infection([[0, 0]])        # -1: nothing can start the spread

Grids can be a few hundred cells on a side and the spread can take thousands of days, so re-scanning the whole grid every day is too slow: aim for O(rows·cols) in total.