fixes to variance computation

[benkeser]

Some bug fixes in how the variance is estimated.

[nhejazi]

Thanks for taking care of this -- really appreciate it. I've temporarily merged this into the branch fix-eif with the intention to eventually merge into master. It seems that some of the code contributed introduces a breaking change here. Would you mind clarifying -- in particular

• The function tmle_eif is meant to be called only in cases of no censoring, so Delta is a variable that is not available when this procedure is called (Delta is only created when C is not uniquely 1). I'm not sure why we would need Delta here?
• I'm not sure why we would create IPC weights in cases when there is no censoring but want to make sure I'm not missing some aspect of the procedure prior to implementing a fix. Is there a reason that we would want to do this?