Implement tilt-to-length (TTL) coupling

Tilt-to-length (TTL) implementation as two contributions (additional optical pathlength) in the ISC (inter-spacecraft) interferometer.

• Uses a linear model in the MOSA total yaw jitter, with user provided coupling coefficients. Two contributions applied to the distant (transmitted) beams and local (received) beams are treated independently, and designated by distant and local tilt-to-length (TTL) couplings. They are symbolically given by

local_ttl = ttl_coeff_local_phi * local_mosa_total_jitter_phi + ttl_coeff_local_eta * local_mosa_total_jitter_eta
distant_ttl = ttl_coeff_distant_phi * distant_mosa_total_jitter_phi + ttl_coeff_distant_eta * distant_mosa_total_jitter_eta

• We remove the propagation of the distant MOSA total jitters in phi and eta, as now the distant TTL couplings are computed "locally" (and available as private data) and added to the propagated beam (similarly to GWs, more computationally efficient because we don't interpolation any new signals).
• Local TTL couplings (also available as private data) are computed locally and added to the distant beam when it is
  • either propagated to the distant MOSA, just like the distant TTL?
  • or when it is propagated to the ISC IFO?
  • or to the local beam when it's propagated to the ISC IFO?
• TTL coefficients are defined as parameters that the user can set in __init__(), either as
  • a 4-tuple of constants (ttl_coeff_local_phis, ttl_coeff_distant_phis, ttl_coeff_local_etas, ttl_coeff_distant_etas) with values compatible with ForEachMOSA init,
  • 'constant' for a fixed set of default values, drawn from uniform distribution [-1.6, 1.6] µm/rad for local coeffs and [-1.65, 1.65] µm/rad for distant coeffs using numpy.random.uniform(), or
  • 'random' to dynamically draw a set of random coefficients from uniform distribution [-1.6, 1.6] µm/rad for local coeffs and [-1.65, 1.65] µm/rad for distant coeffs using numpy.random.uniform().