[Proj] OSGB36

[John Smart]

>> Differences with Ordance Survey results must be attributed to lack of
>> gridinterpolation.
>
> Is that the main thing that OSTN02 adds to all this?
>

OSTN02 is a grid-shifts file with grid resolution = 1km. The shifts are from 
GB National Grid (i.e. a TM projection) on ETRS89 to the same projection on 
OSGB36. Bilinear interpolation is used between grid nodes. The stated 
accuracy is something like 0.1m at 1 sigma. The grid-shifts file has a nice 
feature that it has dummy values outside of a 10km buffer around GB 
coastline. The file also has orthometric height differences (OSGM02) at the 
same grid points.

The 7 parameter is stated as being good for applications needing accuracy 
less good than +/- 5m.

Could be nice if Proj would support this grid shift file, but what is really 
needed (IMHO) is some standardization on such file formats so that a 
plethora of different formats would not be needed. Ntv2 is something of a 
standard but relates to lat/long differences not projection grid.