Linearmodels

* Added System GMM estimator
* Added automatic bandwidth for kernel-based GMM weighting estimators
* Cleaned up HAC estimation across models
* Added `predict` method to IV, Panel and System model to allow out-of-sample
prediction and simplify retrieval of in-sample results
* Fixed small issues with Fama-MacBeth which previously ignored weights

New features:

* System estimation using GMM when there are endogenous variables
* General clean up of HAC estimators
* Many typos fixed

Minor release to address PyPi and documentation issues

New features:

* Three-stage Least Squares (3SLS) estimation of systems of IV equations

Other:

* Make xarray optional
* Small bug and documentation fixes

New features:

System Estimation
* Seemingly Unrelated Regression (SUR) Estimation

Other:
* Remove core dependency on pandas Panel which is scheduled to be removed

New features:

* Linear Asset Pricing Models
* GMM
* 2-step
* Seemingly Unrelated Regression for Traded Factors
* Fama-MacBeth estimator for Panel models
* Autocorrelation (only) robust estimator for Panel models

Other:

* Doc improvements