Kallis out injured meant a no rounder or bowler was playing in his place.
Any other bat of comparable quality out injured meant being replaced by a bat.
The value of Kallis' bowling in the bowling was the degree to which he was better than the bowlers who replaced him. Which admittedly doesn't translate to that much as both he and his replacements took a low number of WPM. And Kallis being medioce isn't a huge upgrade on the poor replacements.
The impact his bowling had on the batting is bigger- it's the degree to which a specialist bat was better with the bat than the no rounders/bowlers. This might 'only' be 15-20 runs or so an innings, but this is still a lot. How would Kallis be rated if he averaged in the 70s with the bat instead of bowling?
So if Kallis had better replacement in case he got injured he would be a better cricketer?
I can think of better arguments for Kallis frankly.
But I think you are also misrepresenting Kallis at that time. His bowling peak from around 95 to early 2000s when he was taking 2 wickets a game also coincided with South Africa being loaded with ARs, batting and bowling. You often had 6, sometimes 7 bowling options, including batting ARs like Cronje and McMillan but also bowling ARs like Symcox, Klusener and Pollock who ensured you already had five bowling options before you even needed to call a batting AR.
So it's just not the case that Kallis was essential, more that he was just that good of a bowler at that time that he could be used frequently and get wickets.
Arguing for impact ahead of stats would be fine. But Akram has 9 too or fewer wicket hauls in losses. His underperformances are contributing to a lot of losses. Even 3 tenfers in losses aren't winning games. Donald had a lot of more moderate hauls in wins. The way Donald took wickets seems to have been better for team results. Are you arguing for better team results, or simply that you like seeing big numbers, even when those big numbers mean small numbers in other games?
You are making a logical jump where you assume moderate hauls for Donald are a winning formula (ignoring that SA was just a better team that win more anyways) whereas if you look at most of those matches there was almost always another bowler who had a standout spell which actually sealed the game and Donald played support.
Shultz in SL in 93
DeVilliers in Aus in 94
Klusener in Ind in 96
Pollock in Pakistan in 97
Boje in Ind in 2000
And if you look outside of those tests, many are drawn games mostly due to SA batting strength that Pakistan didn't have. So in effect you are giving credit to Donald for consistently playing second fiddle to other bowlers in his lineup at key moments. Doesn't sound like ATG standard to me.
The only exception is England and I have already given Donald credit for that.