The actual cluing is a lot more important, and 13ac [deleting 'me' = deleting 'mi'?], 21 ac [Oxford being 'a city' - I think 'a' could have been incorporated in the clue], 25ac [INES is more Portuguese than Spanish], 5d [RASP defined by just 'sound'?] and 22d [too wordy for FA] all have things to worry about.
But I liked 16ac (GHOSTWRITER) and 18d (ROAD RAGE) very much.

I absolutely agree! However…
The main reason I think it would be good to have a sort of ‘guide to Ninas’ (among guides to other quirks associated with the various crossword series) is that the presence of obscure answers can be a clue to a Nina’s presence and I doubt many solvers will have latched onto that. There are lots of simple words to fit the pattern -O-T for example, so why would the setter choose something like HOUT/DOST/TOLT etc? If you examine such oddities you may – perhaps should – be persuaded to check the appropriate rows/columns to see if other letters look like contributing to a hidden string.
As K’sD says, the occasional Nina is well established in the Indy. Other than sometimes forcing in uncommon words, if you spot them they can be helpful; an educated guess about one or two can give you free letters that would otherwise be unchecked.

There’s been much interesting discussion above, and I find myself in the camp that enjoy a good challenge, and the occasional surprise is always welcome, but not at the expense of an overly obscure puzzle, however fair the clues, once solved.

My only real contribution here is the suggestion that the tail of the oriental tiger is indeed very long, financially speaking

My gripe is not with the construction, nor lack of presence, of a nina. It’s with the inclusion in the grid of several words which, as yer average solver, I couldn’t reasonably be expected to have ever come across in a normal lifetime. Or even two. Saggar? Biga? Tael? Yes, all solvable from the wordplay, but I find it unsatisfying to arrive at a string of letters which mean nothing to me, other than they fit the clue. Such words may be buried in Chambers, but if it was stated upfront that some solutions were only to be found in a Klingon dictionary, would that make them acceptable just because they could be derived from the wordplay?

I can accept that sometimes a setter has to use an overly obscure word, and its inclusion at least broadens one’s vocabulary, but several such words in the same grid points to something beyond necessity.

To repeat, I’m nothing more than yer average solver. But average solvers do represent the vast majority of the target audience.