Fix orbit number calculations.

[daniestevez]

This solves two problems with the orbit number calculations:

• When age = 0, the orbit number must match the TLE revnum. However, the expression inside the floor(..) can be give a non-zero value. We need to subtract the result of the floor(..) with age = 0 to account for this.

• In gtk-sat-data.c, the sign of the term involving the argument of perigee and mean anomaly at epoch is wrong. The term should have a positive sign, since it counts as an increasing fraction of a revolution measured from the ascending node.

[csete]

Thanks for the fix @daniestevez.