An interesting and important question—that brings to mind Eddington’s quote: “It is … a good rule not to put overmuch confidence in the observational results that are put forward until they are confirmed by theory.”
Regarding your first question we may refer to Einstein’s isotropic light-speed assumption (postulate, or stipulation). Anisotropic light-speed, within which c=constant is a special case, has been promoted in recent years (e.g., Rizzi, et.al, “Synchronization Gauges and the Principles of Special Relativity”, 2008). By itself anisotropic light-speed is of no scientific (empirically meaningful) consequence, but within expanding space—i.e., Hubble expansion? Here we may consider two alternatives: (1) No theoretical effect, and so the Lambda-CDM model passes another test; and (2) There is a (theoretical) effect, however small, which either challenges the supernova measurements and conclusions or leads to a solution.
So it appears high-redshift Type-Ia supernova measurements comprise the first step, with at least two steps remaining (including a possible explanation for cosmic acceleration).
Where is Wiltshire's timescape cosmology. It argues that the recent acceleration is an effect caused by the inhomogeneity of the universe, combined with the fact that we are located not in a void, but on a border between the voids, thus, a high density region. Expansion is faster in a low density part, and what we see is light which has been, in recent time, traveled through such voids which have been expanding faster than our local environment.
This explanation also explains why the "accelerating universe" is only a recent effect. The explanation is that the inhomogeneity was ignorable in the past, and the effect becomes visible only once the universe becomes sufficiently inhomogeneous locally. Which is the correct timing.