Are accelerated reference frames as valid / real as inertial frames, or are accelerated frames a convenience of integrating across a continuum of inertial frames?
For instance, it makes sense to consider a passing relativistic observer for some experiment done on earth, if only to imagine what they would observe on earth due to relativistic effects. It then makes sense to consider any and all possible relative velocities / momenta, meaning that any and all inertial frames must be considered in that they all should find a way to agree on the events taking place on earth.
Does the same property hold for non-inertial frames? Eg, does it make sense to, or is there a need to consider every accelerating frame of reference such that they would agree on the outcome of an observed experiment on earth?
If so what are the consequences of acceleration horizons giving rise to a set of accelerated frames that, due to distance and acceleration, are unable to observe an experiment on earth until they stop accelerating, while another set of accelerating frames can?