"The question is about coordinating 'ratings' and 'performances' in races to deliver a rating for each race.
Can this be done?"
To measure performance and make that rating transferrable across races/meets, you need a good template, and ability to measure any deviations from that template. If you want to "deliver a rating for each race", something workable across the board, you have to leave behind conventional methods of race analysis.
Imagine if the world of Race Analysis were an Art Gallery, then all the pieces on display would be the same, or nearly the same. And this Art Gallery of ours has, since the 1970's, seen little change.
I am not saying that the "conventional methods of race analysis" are of no use, just that they are inadequate.