Consolidate other ML libraries in Dask-ML

#241 is bringing dask-searchcv into the dask/dask-ml repo.

Should we do the same for

- dask-glm
- dask-tensorflow
- dask-xgboost

My motivations for merging in dask-searchcv were

1. dask-ml is set up to run against sklearn master
2. dask-ml sees more activity these days, so we're running CI more often and will catch breaks sooner.
3. It's less work to release one package instead of two :)

All of these apply equally well to these other repos.

Of the three repos, dask-glm has the most distinct scope (focus on optimization), but I assume most of its use comes via Dask-ML

cc @jakirkham