<<I'm chary of experts determining what sources are reliable, as Carcharoth suggests.>>
Experts do not determine what sources are reliable. Consensus does.
<< There are two meanings for "reliability." Reliability in RS, I claim, depends solely on the publisher, and reliability in this sense is about notability, and certainly not about reliability in the ordinary sense, that we could assume that the material is "true." If it's in independently published source, it's reliably sourced. Sure, there are gray areas.>>
No. Reliability as we use it in WP:RS depends on an author and the publisher's editor being known as producing resources which accurately reflect their own underlying sources. It is not about the notability of the publisher, because in that case a *famous* publisher like Ivana Trump Enterprises could produce complete dreck and be called reliable. WP:RS is a combination effect arising from the interface of author-publisher-source. It's not dependent on one of these alone.
<<If we accept that fact in reliable source -- or "asserted fact", to be precise about what can be verified -- is usable in the project, my view is that RS establishes notability and that, therefore, the fact belongs somewhere in the project, it should not be excluded because someone, expert or not, claims that, say, the author is biased. Rather, if that impeaching claim can be backed, itself, by reliable source, we would provide both, and the original "fact" would be stated with attribution, "according to ..." and probably likewise the rebuttal. Even if there is no impeaching claim in reliable source, it is within the sovereignty of local consensus to include attribution where it will broaden consensus.>>
RS does not establish notability. Example among a few hundred other deeds that Bogislaw I did, he also loved partridge pie. It's sourced reliably, that doesn't necessarily means it's notable. Things which are notable should be important, interesting, standard, curious, odd, startling and so on, not mundane, bland and trivial regardless of the source.
Next IF you have an example (I'm sure you do, you're just too shy to tell us its Britney Spears or Scientology or both) where a single expert has decided that a source is biased and is therefore blocking that article, then tell us so we can go beat him or her up.
If you don't let's just say that a single editor should not be able to OWN an article nor the use of a source or it's disuse. Community consensus still prevails and there is an WP:RSN reliable sources noticeboard where you can bring that forth and establish a community posse to take out the varmit.
IF however your source states something that *the vast majority* (and I characterize this as meaning "me") would say "that claim is outrageous" then you have to recognize that fact and abide by the community standard view. So "apple pie with whipped cream has been shown to cure some types of cancer"... no. "Space Aliens tend to like to visit San Diego more than Los Angeles".... no.
"Some medical studies show a correlation between high blood pressure and doughnut-eating" (Source 1) "while other studies have found no such correlation" (Source 2).
There is a point at which a claim falls off the face of the Earth because most of the community is rolling around on the floor laughing.
So bring your example, mr shyness.
Will Johnson