In theory this allows GOES-R line/column/xFAC/xOFF to be used thorugh goes_nav(),
aka cgms_nav() with additional parameters

needs a unit test!

#17 fix - NaNs in cgms_x in cases where S0110 not involved, needed to use meta.begin_line to query inside available lines

- extents of scene (line count) now includes any missing lines at start of scene
-- use start and end lines to find true available data extent, if needed
- slightly better error reports
- ERFA library updated to include 20150630T235960 leap second

note that comma-separated donor list could be used if not all scenes have B16

Added new version of erfa from
https://github.com/liberfa/erfa
Source merged with:
python ../erfa-fetch/source_flattener.py src -n erfa

Preliminary code to generate proj4 string for CGMS03 projection,
with the intent of rudimentary but corrected GeoTIFF using GDAL

adds shape (maximum available line/column) vs extents (subset of shape known to actually contain data)
metadata API now returns shape (end_line) instead of extents (end_line - begin_line)
ticket #1 potential resolution

nan("") instead of nan(0)
execlp warning addressed but probably overkill

also updated table in hsd2nc used for downsampling
Note that *Coords query in libHimawari was unaffected since it does calculation of
CGMS coords internally, using the from-file values instead of the HimawariNavigation
query.

- already had obi-wan fix for column storage coordinate calculation
- had to move rotational correction to after the correction table shift