check min/max attributes against data range

[leilabbb]

from Travis Miles email exchange.

Example of an encounter of data quality issue with glider - ng1260:

An initial investigation shows that:

1. The valid_max (= 10) and valid_min (= 0) for conductivity do not reflect the data range (> 10), so the data reader masked the data values. As a result, the data were not evaluated and the flag was set to 2 (= not evaluated).

2. The unit is set to S m-1 in the profile files, but in this case it was likely intended to be mS/cm, which would have explained the data range. Reporting the data in the correct units would have been converted during the mapping of the standard_name/units pair built into the GDAC software.

The global attribute dac_qc_comment in the individual profile files lists the QC issue with the variable conductivity (dac_qc_comment = conductivity is an array of masked values). If it were not for the masked values, the GDAC gross range test would have caught the failing data (> 9) and set the flag to 4.

Solution:

• Add a test to check the min and max attribute values against the data range.
• Add more details to dac_qc_comment to report the error encountered.
• Test setting the flag to 4 (= bad) if the data are outside the data range.