Future code improvements

[jburba]

A non-exhaustive, unordered list of changes we would like/need to make to the code in this repo:

• Accounting for time correlations (Mike W. circulated a PDF at some point with some math on this...)
• Support for flags which are not uniform in time (we need to recalculate Ninv using the flags for each time step which may or may not require recomputing the entire GCR left-hand-side operator)
• Support for polynomial bases for the FG model?
• Including a monopole term in the EoR model (this arose from testing H4C EoR validation sky model simulations with hydra-pspec, the non-zero monopole in the EoR sky model in this case resulted in EoR visibilities which are not zero-mean per frequency and the current EoR model cannot account for this)
• Support for systematics modelling (led by Sohini D.)
• Support for time-varying noise covariance matrices?