Replace 'me' with 'mi' terms

[paul-buerkner]

With the latest developement of missing value 'mi' terms, which now can handle measurement error and missingess at the same time, measurement error 'me' terms are now merely a special case of 'mi' terms. Further, 'me' implicitely contribute the meaurement model without this being appropriately accounted for in the modeling syntax or post-processing. Currently, me terms have some advantages for the post-prcoessing with new data, but these advanatages are only caused by the implementation and do not refer to any actual statistical cause.

Once the post-processing of 'mi' has been improved, which basically means allowing for sequential evaluation of predictions of multiple responses, 'me' can be deprecated and eventually removed.

[paul-buerkner]

mi terms are now very close to fully support everything that is offered by me terms in addition to all the other flexibility and consistency that comes with mi. I have decided to soft deprecate me terms, but I don't think it would make sense to actually deprecate it (with the thread to remove it eventually) but rather keep it around for the time being.