1) The terse answer is, if your heuristic is not admissible, you will (possibly) get a non-optimal result. I think you knew this. For the intuition, recall the definition of an admissible heuristic: It is a heuristic which is never more pessimistic than reality. (We usually say, "it is always optimistic," since if you had a heuristic that was neither optimistic nor pessimistic, you'd basically have your answer already.) If your heuristic is pessimistic in some places, then it ends up avoiding the best choices.
As for scaling up and scaling down the heuristics as per your question, remember, you are only scaling up the heuristic part of your formula, not the sunk cost part of the formula. If you could scale them up exactly the same, you couldn't see the difference, but you can't always do that. Even in your example, the additive bit you tacked on spoils it.
2-3) It's not clear to me what you mean by "solving" pacman. If it's anything more complex than finding a shortest path to eat all the dots in an empty grid, you're getting well beyond the scope of A*, I think. Even then, A* would not be my tool of choice.