Indeed I think we are in agreement on pretty much all counts. I too agree that it is useful to distinguish between different levels of NS claims - on the one hand it is clear that humans are exposed to NS (e.g., non-viable zygotic genomes or in response to epidemics) but on the other hand it is imprecise to not have a rigorous mathematical definition of NS’s survival or reproductive fitness.
FOXP2 is a good example of a gene that illustrates these modeling difficulties. This is one of the genes in which both Homo Sapiens and Neanderthal differ from other primates and it is thought to be significant even though the difference is in only 2 amino acids out of 715 (T303N and N325S). A major reason why it’s thought to be significant is that it was shown in Lai et al, Nature 2001 that even just one mutation (R553H) on this gene in the KE family can result in significant physical (rigid lower half of the face), cognitive (lower IQ) and language impairments (difficulty pronouncing words). The reasoning then goes that the human version of the gene could have resulted in substantial gains in language and cognition and thus be an important factor in the biological underpinning of human phenotypes. But how can this be tested? Konopka et al, Nature 2009 approached this problem by substituting the chimpanzee version of FOXP2 for the human version in human cell lines and measuring the immediate effects in terms of gene expression and observed molecular/cellular phenotypes (which I suppose is not unlike the other references you were alluding to in your post). However, it is easy to see that the ultimate organism-level phenotypic impact of even just one mutation on just one gene can be very difficult to determine - it is simple to measure neurons’ mobility, axon growth, synapses, etc but those measurements are orders of magnitude removed from the ultimate cognitive impairments resulting from just this one mutation. To make things even more complicated for NS, it is not at all clear that the cognitive impairments resulted in any decrease in reproductive fitness - according to Lai et al, Nature 2001, every single member of the KE family with the mutated version of the gene actually had more children than the one couple that was homozygous for the non-mutated version of the gene.
Circling back to the question on whether NS should be a law or a hypothesis, it would seem to me that it can be both but we must be clearer as to what is being claimed. While it would seem simple and uncontroversial to claim a law-type sense in that genotypes can fall bellow the survivability bar and be “naturally selected out” (e.g., non-viable zygotic genomes or in response to epidemics), it is also true that a lot is missing in being able to predict the phenotypic fitness of an organism so we can’t yet show whether NS’s reproductive fitness suffices to determine which genotypes survive and which don’t (more of a hypothesis in that sense, even though it’s overwhelmingly more supported by experimental data than any other approach).
A possible analogy that may be helpful here is that Copernicus and Galileo were right that Earth orbited the Sun way before Newton, Kepler and later Einstein and others figured out a lot more details on the math, planetary orbits and the nature of gravity itself. Earlier models were imperfect and did not explain all the observations but they were still accepted as valid theories and laws of physics even though they were “work in progress” for centuries. As with the rest of science, the key aspect of these models was that they made testable/falsifiable predictions that were mostly confirmed by experimental observations, similarly to what we see for ND/NS if we are careful in defining the scope of the claims.
All in all, it may well be that the major theological implications of genomics/evolution at this stage are the concepts of common ancestry and minimum populations sizes (as it relates to literal readings of Adam and Eve or Noah’s genealogies). Compared to these, it would seem to me that the impact of God creating and guiding using ND/NS versus using ID or teleology is relatively minimal, but that may be just me.
As this may be our last exchange of the year 
, I wish you and yours a blessed and happy 2016.