If dates don't line up exactly at the minute mark, the frame gets filled with np.nans
now frame takes closest record that is < than the minute mark and sets
that record into the file

Some variables weren't in frame and qc were marking them as 'good'
Now they are marked as missing as desired

need to reorganize headers in the future

schema contains valid_min and max values
level_b1's nc.py now implements those values into the netCDF header:

every five seconds, a new rolling 2 minute average is calculated
every minute, only 1 average is stored in the file's memory

averages now start at future minute - 1 to future minute

averages all vars now except wind_speed and wind_direction
Haven't added gust yet