number of color levels chosen varies with scaling

[ACManke]

Longtime user Serena Illig reports in this message https://www.pmel.noaa.gov/maillists/tmap/ferret_users/fu_2019/msg01137.html
that the same data, simply scaled by powers of 10, and plotted using color levels also scaled in the same way, results in 16 color levels in some instances and only 15 color levels for others.

I think that this is related to Ferret issue 1877

It is not the single-precision vs double-precision problem we sometimes see with PPLUS code, but is related to the fix for 1877.

[ACManke]

fixed with changes to tm_fpeq_eps.F, tm_fpeq_eps_sngl.F and levset.F