Product codes assume IMOS QC flag scheme but don't check for it

[mhidas]

The the aggregated product includes QC flags in the output file, while the hourly product excludes data values not flagged as "good" or "prodbably good". Both of these make the implicit assumption that the input file uses the "IMOS standard flags".

More specifically,

• All _quality_control variables in the aggregated product output files are given the attributes:

  "flag_meanings": "No_QC_performed Good_data Probably_good_data Bad_data_that_are_potentially_correctable Bad_data Value_changed Not_used Not_used Not_used Missing_value",
  "flag_values": "0, 1, 2, 3, 4, 5, 6, 7, 8, 9",

• In the hourly codes, values are only included if their QC flag is 1 or 2 (without checking the "flag_meanings" attribute in each input file).

The assumption is correct for the vast majority of IMOS moorings data, except for the Acidification Moorings, which use the WOCE QC flags:

flag_values = 2b, 3b, 4b ;
flag_meanings = "good questionable bad" ;

Luckily this is sort of compatible, in that selecting only flags "1 or 2" gets us only the good data. However, for the aggregated product, re-labelling these flags as if they were from the IMOS scheme is misleading. In effect all the "good" data (flag 2) will now look like it's only "probably_good_data".