Fermi -> Wang Ganchang -> Harrison, Kruse & McGuire took a while. I wouldn't expect a quick detection of something with the properties of a WIMP, as I would expect any such particle to be even harder to detect than a neutrino, especially if it doesn't feel the weak nuclear force.
I don't know why you think that dark energy has failed to show up in any conceivable form -- how do you explain the cosmological redshift without it? Dark energy in its simplest form is just the cosmological constant, and can be an inertial effect.
In any case, even if the concordance cosmology is simply wrong, that does not mean that GR is incorrect as much as we are wrong about the mechanisms that generate the metric (Afshordi, one of the authors of the paper at the top, has proposed non-universal coupling to the single metric of GR), or alternatively we are wrong about the way we choose and stitch together metrics (i.e., we're misusing GR in a way that introduces serious errors at long length scales).
Do you really think there are working scientists who think that there's nothing more to discover? Conversely, are there many who deny that the huge preponderance of evidence we have so far favours the Standard Model and General Relativity? Even if we "demote" SM and GR to effective field theories, the effective limit of each is very nearly everywhere readily accessible, isn't it?
Buying the BH singularity would be, I think, a pretty extreme position. Every viable post-GR effort I know about is to some extent focused on abolishing singularities somehow. (You could alternatively keep them always hidden and resolve things like BH thermodynamics; if you always keep information locked up in another region of spacetime -- behind a horizon -- there's no information loss problem to consider, and you can "cut" singularities out of the manifold recovering everywhere-smoothness. But black holes might evaporate completely in the far de Sitter like future.)
Not buying an event horizon in a system of local physics with a maximum propagation of local state from one point to another seems even more extreme. The existence of a maximum local speed -- whatever it is, it could be much faster than light -- sets the slope of a nonempty open convex cone of tangent vectors (a causal cone at each point for the field-values at that point) which in turn lets us fix a first order quasilinear system of PDEs admitting a hyperbolization, and in that you can always find an observer that sees an event horizon.
The formation of the BH creates a dynamical spacetime with an acceleration between observers before and after the collapse, and that alone is sufficient to produce an event horizon.
Abolishing 'c' (as a general free parameter defined at every point; the definition can even vary by location in spacetime) seems a lot harder to swallow than abolishing event horizons. If you accept 'c', then while Schwarzschild event horizons is pretty easy (nonzero J or not-always-zero Q at all physical compact dense objects, for example), abolishing all event horizons requires a lot of contortions to avoid immediate conflict with local experiment, much less astrophysical observation.