The logic instrument.apply_shift is now in fixed_shift_numpy.AdaptiveShiftNumpy

This emulates the previous behavior during the refactoring but needs to be improved later.
It allows invalid points in shifted data also near right boundary.
The ranging unit test that was failing with stricter BC now passes again.

This adds a new method apply_shift, which uses the daskified time-shifting. It also changes
the interpolation config parameter, adding minimum and maximum delay to the tuple. This delay refers
to inter-spacecraft delays. The new methods are not applied to other delays yet, which are still using
the old interpolator code that is still in place.

There are now two dask-able Lagrange interpolators to chose from, a new implementation (method "lagrange")
and one using dsp.timeshift internally ("lagrange_dsp"). For both, there are now separate interpolator
interfaces for fixed and dynamic delay. The code to chose fixed or dynamic interpolator based on argument
type is now in the new instrument.apply_shift() method.

There are new unit tests for the time shift, also testing old and new implementation against each other.

All but one of the original integration tests pass. The fail occurs in test_ranging.py for test
test_prn_ambiguity_with_esa_orbits. The cause is that the sampling rate is chosen so low that the
interpolation stencil extends beyond the right boundary, triggering an error by the stricter
checks of the new routines.

Now a positive shift generally means a shift to the right, i.e. a delay is a negative shift.

Add relative shift location selection to interpolator interface. This will allow wrapping dsp.timeshift.

This includes a rewrite of the Lagrange interpolator using a new algorithm based on FIR filters.