Update BUFR document's reference to heights

[Ian Culverwell]

ROPP-10.0 beta reviewers state:

During the IROWG 2019 meeting of the NWP group, one commercial provider asked
for the BUFR format description to explicitly state that heights shall be given
with respect to the WGS86 ellipsoid.

I think they mean WGS84.

[Ian Culverwell]

I asked Chris M for clarification. He said:

What I just found in my notes is that Vladimir Irisov from 
Spire suggested to explicitly state that the local radius of 
curvature should be calculated as local radius of curvature of 
the reference ellipsoid (WGS84, I assume).

On the heights that are available in the BUFR files - yes, I 
agree that they should all refer to the geoid, hence EGM96. Is 
that somewhere mentioned in the BUFR documentation? If not, 
would it make sense to add a footnote or something?

Agreed. Entries 35, 37, N1+2, N2+2 and N3+2 of Table 5 of the BUFR document have accordingly been changed at r6505.