Improve sd flagging

[kathryn-willi]

I think that part of the flagging issue is that the flag ends up being centered, or on the left or right of problematic data, so it doesn't look like it's catching all that it should. I think that if we think about how each of these flags are working, we can define rules about how to add neighboring flags (i.e., broadening the flagging window)

[juandlt-csu]

I think that the updated sd flagging works great with the new method that we are using (sd flag within the 10-99 quantiles). I think something that we could add is a distinction between flags that exceed 3sds and 2 sds.