1. Is this game worth it, or should I focus on finding a better game somewhere else?
It depends on what "worth it" means. This is not a good game by most standards. You don't say whether it is an S17 game or an H17 game. I will assume it is an S17 game, that is, a game where the dealer stands on all 17's, including soft 17. Based on a sim from Norm Wattenberger's CVCX software, this game penetrated to one deck, where 75% of the cards are dealt, earns 1.77 units per one hundred hands played, with a per hand SD of 3.78 per hand, using a 1-8 spread as described below. A 1-12 spread has an EV of 2.89 units per hour and a per hand SD of 5.00. This is using the KO Preferred style of play, which I would strongly recommend over KO Rookie.
2. What bet spread is optimal for this game?
If you are using a 1-8 spread, try betting 1 unit at -2 or less, two units at -1, 3 units at 0, 6 units at +1 and 8 units at +2 or greater. If you are using a 1-12 spread, bet 1 unit at -2 or less, 2 units at -1, 4 units at 0, 6 units at +1, 8 units at +2, 11 units at +3 and 12 units at +4 or greater.
3. I started counting with Hi-Lo, but have since switched to the KO rookie count. I am assuming the IRC should be -12, but what is the proper key count for 4 deck play? I have been using -2. Is this correct?
As noted above, you should move on to the KO Preferred system. It is not a big stretch, since there are only a few strategy plays that you need to memorize. Using an IRC of -12, which is what you wish to do if you want to stick to the strategy indices in the book, then your Key Count is -1 and your pivot is still +4. This is based on a post made by Ken Fuchs on T-Hopper's website back in December of 1997. If you are using the KO Preferred, you would eliminate Category C plays altogether, just as you would in six- or eight-deck games.
If the game penetration is less than 75%, it should be avoided altogether, unless you can wong in and out with ease.