Playing with R:
arid <- function(toc) { sprintf("%08x-%08x",sum(toc),sum(1:length(toc) * toc)) }
arid(c(183, 22015, 42950, 60680, 79713, 100643, 119240, 138148, 156143, 175288, 193653) - 150)
arid(c(183, 22015, 42950, 60680, 79713, 100643, 119240, 138148, 156143, 175288, 193653) - 150 - 8)
[1] "0010961e-0083c868"
[1] "001095c6-0083c658"
You can play with it here: http://www.compileonline.com/execute_r_online.php
Just paste the code snippet, click "Execute Script" and observe the first part of AccurateRip Id appear on the right.
Looks like CUETools somehow assumed that pregap is 8 frames shorter. I guess it got the wrong impression from a log file or CTDB. This theory can be tested by unchecking CTDB verification and/or moving a log file to a different folder.
33 frames pregap being mistaken for 25 frames pregap is also consistent with mistaking frames for 1/100 of a second: 0.33 of a second is roughly 25 frames. Depending on EAC version (and probably settings), log file can contain gap lengths in either frames or 1/100th of a second, and CUETools can only guess how to interpret that.