Clairemont,
Regarding my second reference, scroll down as far as you can on that screen. You'll seen a heading with a purple background that says: "Population Diversity (Alleles in RefSNP orientation). See additional population frequency from 1000Genome...". Underneath that is another heading of: Sample Ascertainment Genotypes Allele, and underneath that is another subheading.
In the subheading, Population means the ethnicity of the population studied. So EUR is European, AFR is African, and I'm not sure of the others. Then you have Chrom. Sample Count. That is how large each sample was. In this case, under Alleles, if anyone in each particular sample had an A allele, part of the bar graph would be green. If anyone had a G allele, that part of the bar graph would be gray. Above each bar graph are numbers representing the portion of population that had an A allele and the portion that had a G allele. I misspoke above when I said no A alleles were found, because I was just looking at the color of the graph and didn't look at the numbers. The number of A alleles found in the EAS was too small to show up on the bar.
Actually, for the EAS population studied for rs72554664, .0050 had an A allele, and .9950 had a G allele. So 99.5% of that population had a G allele. For the aggregated populations, .0004 had an A allele and .9996 had a G allele.
I also looked up the population diversity for the other 2 snps you listed on dbSNP, which is the shorthand name for that second reference site. For rs5030870, with an aggregated population sample of 121,410 chromosomes, there were no A alleles found. For rs35972742, the sample size is 2 chromosomes, identified as Watson, which I presume is the famous Dr. Watson of double helix fame. He was GG.
Based on the limited information that shows on dbSNP, it seems these snps have not been studied much. I don't know why you think your versions are likely fairly rare. From what I see on dbSNP, I would conclude that you have the most common versions of these 3 snps. If you have symptoms that you think might be due to G6PD, it might be from the other G6PD snps that are discussed on the OMIM link I gave. My v.3 23andMe results lists 59 different G6PD snps.
When you look at your raw data on 23andMe, if you look at the far left side of the line for each snp, you'll see a little plus sign in a box. If you click on the + sign one time, a drop-down menu appears, and one of the choices in that menu is dbSNP lookup. Clicking on dbSNP is an easy way to check out how common your alleles are. SNPedia also has a more simplified chart of how often allele combinations appear in various population groups. I also like to click on Google Scholar (SNP) and Google Scholar(Gene) to get more information.
To look at the population diversity for a well studied gene, MTHFR C677T, look at
http://www.ncbi.nlm.nih.gov/SNP/snp_ref.cgi?rs=1801133.
I don't know of any guide for how to use these gene interpretation tools. I agree there is a need for one. I think someone posted a tutorial about genes on the blogs.