Makes sense. I agree wiki is probably the wrong place. Maybe expanding the inline documentation in the source code. I don't really think this kind of code is useful to most applications but there is a lot of interesting stats which can be pulled from the UXTO. Distribution of output age, distribution of outputs by script type, distribution of output values. One could build the UXTO from the raw blocks but since the chainstate has that data (albeit in an somewhat opaque format) it seems a smarter place to pull that out.
FWIW, the gettxoutsetinfo rpc iterates over the utxo set, in the past when I've wanted this data I've just added some instrumentation in that function to dump it out. Even if you're not very experienced with C++ it shouldn't be to hard to emulate the rest of the code and print it out... this might be easier (and also more reliable across versions) than trying to read the data.
That is a good option. I would love a dumptxoutset RPC command. Still if was an interesting project and I learned a couple things so that is always good. I never considered that for uncompressed keys in the output set you don't need to record all 65 bytes. could just record the x and recompute the y as it is done for compressed keys. I new tx version could save 32 bytes per input (Pay2PubKeyHash) or 32 bytes in the output (Pay2PubKey) by recording the PubKey in compressed format while still supporting uncompressed PubKeys.