Thanks Pascal.

Fix #566

The test required EOP data, but did not configure the data path.

Because of the addition of a static field for the GLONASS_EPOCH in AbsoluteDate
leap seconds had to be loaded to use AbsoluteDate. This commit restores the
previous behavior where AbsoluteDate can be used without loading leap seconds
first.

Thanks to Giuseppe Cavallaro for the translations.

This is mainly useful when sampling areas of interest that cover the
pole as the pole is singular for classical aiming directions (constant
azimuth or along track).

Thanks to Pascal for the very clever idea of diverting singularity!

Fixes #547

As per Brouwer's hairy ball theorem, any vector field on the 2-sphere
has at least one zero. This implies that any implementation of
TileAiming interface has at least one points where the aiming direction
cannot be computed. The most typical example is aiming always towards
North pole, for which both poles are singular points.

Fixes issue #558

Fixes #559

Neither ContinuousSelector nor BurstSelector can be reused across
several schedulers during measurements generation because they store the
last selected dates.

Fixes #557.

Improved test coverage for FieldSpacecraftState and SpacecraftState.

Fixed Javadoc for AbsolutePVCoordinates.

Improved test coverage for InertialForces.

Addition of new tests for GNSS propagation and tropospheric models.
Moreover, addition of UT for classes related to the propagation in non
inertial frame.