I wouldn't agree with that post - the probabilities and number of clusters imply we'd expect a bullet cluster. In fact, we observe other similar lower-mass systems.

As an astronomer in the field of clusters, I don't know anyone who takes MOND-like theories seriously. Any non dark matter explanation for the bullet cluster would have to be extremely contrived.

I doubt she thinks MOND is a candidate for a family of fundamental theories (MOND itself is not relativistic; fixes for that -- at least where the result is a metric theory -- tend to cause geometry problems at solar system and cosmological length scales, and relic radiation is also extremely sensitive to geometry; and why would an extended metric theory have a better answer to perturbative non-renormalizability?).

However, there's lots of scope for taking an unphysical and even incomplete theory seriously [this supergravity researcher puts it well: https://arstechnica.com/science/2013/05/earning-a-phd-by-stu... ]

Relativistic approaches to MOND might eventually raise interesting questions about the EFT, and those are what a QG phenomenologist lives for. So it's easy to see why she'd take MOND seriously without taking it too seriously.

(As an example, Verlinde's recent work is partly provoked by MOND, and it's also nice to see a string theorist step outside of the AdS comfort zone and confront what we see in our sky).

  the probabilities and number of clusters imply we'd expect 
  a bullet cluster

Not according to the references in that article, which put the probability of seeing such a cluster at 10E-4 to 10E-6. And Sabine isn't one to conveniently forget to multiply that probability with the number of clusters: that's already factored in.

The Thompson et al paper linked is precisely a probability for pairs of clusters, so needs to be scaled (see comments on article). There's a nice comment in the article by Peter Erwin giving an overview of a list of recent papers:

  Hayashi & White 2006 -- BC is apparently consistent with LCDM.
  Farrar & Rosen 2007 -- BC is apparently not consistent with LCDM.
  Nusser 2008 -- BC possibly not consistent with LCDM.
  Angus & McGaugh 2008 -- BC is more consistent with MOND cosmology than LCDM.
  Llinares et al. 2009 -- BC is more consistent with MOND cosmology than LCDM.
  Lee & Komatsu 2010 -- BC is apparently not consistent with LCDM.
  Forero-Romano et al. 2010 -- BC is apparently consistent with LCDM.
  Thompson & Nagamine 2012 -- BC is apparently not consistent with LCDM.
  Watson et al. 2014 -- BC is apparently consistent with LCDM.
  Krajlic & Sarkar 2015 -- BC is apparently consistent with LCDM.
  Bouillot et al. 2015 -- BC is apparently consistent with LCDM.
  Thompson et al. 2015 -- BC is apparently consistent with LCDM.

LCDM is Lambda CDM, the prevailing cosmological model. BC = Bullet Cluster. The blog post is very selective in the papers discussed.