add "_Encoding" attribute to time_iso history file variable

[jswhit]

The 'time_iso' variable is an array of characters. The "_Encoding" attribute allows netcdf clients (like python) to automatically convert the array of characters to a fixed-length string (see Unidata/netcdf4-python#654 (comment)).

Before this patch,

>>> nc = Dataset('test.nc')
>>> nc['time_iso'][:]
masked_array(data=[[b'2', b'0', b'2', b'1', b'-', b'0', b'8', b'-', b'3',
                    b'0', b'T', b'0', b'0', b':', b'0', b'0', b':', b'0',
                    b'0', b'Z']],
             mask=False,
       fill_value=b'N/A',
            dtype='|S1')
>>> 

After this patch,

>>> nc['time_iso'][:]
array(['2016-01-04T06:00:00Z'], dtype='<U20')
>>>

See also Unidata/netcdf-c#402

Since this patch is tiny it might make sense to merge it into another larger update.

[junwang-noaa]

@jswhit Will this code update change results?

[jswhit]

@junwang-noaa no it will not change results - it just adds another attribute to the time_iso variable.

[DusanJovic-NOAA]

Merged via #550