[Feature]: Mean recruitment adjustment

Describe the solution you would like.

see discussion in #593
When using recdev option 1, the dev_vector with tuned bias adjustment forces the arithmetic mean recruitment of the time series to be consistent with the mean from the spawner-recruitment relationship.
This is because the spawner-recruitment relationship is used to produce MSY; MSY is not based on the time series of estimated recruitments.
With the transition to recommending that recdevs be based on recdev option = 2, which is not zero-centered, there is a concern that the MSY calculations will not be consistent with the mean of the recruitments.
It seems that MSY should remain based on the SRR, as adjusted by the mean deviation of the recdevs.
Solution 1: add a penalty to mean recdev to improve degree to which it is 0.0
Solution 2: base the benchmark quantities on log(R0) - mean(dev)

Describe alternatives you have considered

Rely on the sigma penalty to keep mean recruitment close enough to zero. However, it is clear that his does not always happen. A change to log(R0) can be perfectly offset by a drift in mean recruitment.

Statistical validity, if applicable

No response

Describe if this is needed for a management application

definitely important to not allow a know bias in this calculation

Additional context

No response