All records get a different draw from EPS

[kylebaron]

Previous behavior was to assign a new EPS only when time was advancing in the data set within individual. Now, we assign a new EPS every record.

There was some linting that happened too ... not sure if this came from Rstudio or not. But let's leave it as is.

[kyleam]

• The code change makes sense and looks straightforward (though I don't have a solid grasp of all the moving parts).

• I confirmed the new test fails if I revert the code change.

• I confirmed the internal cases that prompted this PR are resolved by this change (i.e. we no longer need to force any covariance matrices to be positive definite when calculating npde).

• Reading through the internal discussion, it sounds like there is consensus about this change in behavior.

• As far as I understand, this can be viewed as a pure bug fix. In that case, I think unconditionally changing the behavior along with a NEWS item is sufficient, as opposed to doing something more involved to retain the current behavior.

[timwaterhouse]

Looks good. Agree with @kyleam that the change along with the NEWS item is sufficient.