About 1 -- I heard this before - this is fine with me -
An implementation Q: should we also NULL pad the gaps on the raw data (15-min) or just the derived stats (Daily or Monthly stats)?
We can tackle at the same time doing checks on stats to purge biased values, although I dont mind decoupling both actions. Do we need to run this by all, or we just do this for BB and Mets and that's that?
I must say that some software helps better than others, but I'd probably also use matlab, and i feel the pain too -- actually, the new versions of matlab may have gotten more savvy about handling dates, but I did not use the stack intensively for what it feels like a decade.
> In my view, mean daily should output a date even when data is missing. This way you can easily compare e.g. bb data and met data using one date column.