Frequency error or drift in pwl_cos

[sgherbst]

I noticed that the frequency of the sine wave produced by pwl_cos either has a slight error, or is drifting over time. I observed this issue in the DragonPHY project when working with ADC performance metrics like ENOB. A drift or frequency error could be due to many things (including simulator precision), but it's worth looking at the way phase is accumulated in the cosine model to make sure that is not the culprit.