So I revisit this question every once in a while, constantly starting an answer with some resources, and then deleting it. Long story short: I think that there are several places where such items could be found, though I think "good" can be hard to define (as it is so situational)!
First off, here is a link that provides a few options for most of the contexts you described. I like this particular list as it describes multiple options (though sometimes there is little to distinguish) for each class of question. Perhaps this is a resource that could be used.
Let me emphasize that, in my experience, demographic questions are very context specific. For example: why are you using the questions? Is it for sampling weighting purposes, where you wish the sample to match population values? If so, then the "correct" demographics to use would be those that best match the provided population information. Is it to be included in analysis of some kind? If so, what are your research questions? A question asking the relationship between age and weight would ask for very different information than a question focused on the differential functioning of a psych evaluation between those college aged (18-22) and elderly (65+). In the first, a demographic question with a write in answer would likely be most appropriate (albeit perhaps biased), while in the second categories would serve best (with the categories being dependent on the question: so 18-22 and 65+ should be two of the categories).
As I believe that these are so connected to the situation, I'll also recommend ICPSR as a place where questions from multiple surveys can be looked at side by side. I've used this to compare many different versions of demographic questions to see which one fits the situation best (though again, I wouldn't say any are "good" out of context). This type of feature, that which compares multiple versions of the same class of question, may the best recommendation that I could give. It gives multiple options, while leaving room for professional judgement in the context of the use of the measurement instrument.
A final thought. If there is not a particular purpose for the use of demographic information, then strongly consider not including such questions in your measurement. Often researchers wish to all sorts of information (as "what if I need it later?"), but this can have some adverse effects: it increases fatigue, increases the time it takes for participants, and can increase the ability to identify individuals based off of the information (potentially changing it from anonymous to confidential, or even open!). It is not hard to imagine a researcher studying teachers who wishes to know a whole series of demographics, in addition to the grade, school, and subject an individual teaches. The degree of specificity from these demographics can make participants easily identifiable. All this to say: if you don't need them, don't use them!
In closing, I recommend only using demographic questions that are necessary for your research (this includes exploratory research). If used for a specific purpose, I would challenge that the purpose often dictates the structure of the demographic question to be used (as can the population). As the purpose and population can be so different from situation to situation, I suggest the comparisons of multiple versions of the class(es) of demographic item(s) that you need, and selecting from them the one that best fulfills your purpose. For this goal, I suggest ICPSR as a great place to look at surveys who speak to a wide audience (and are well designed).