I was naturally skeptical of the punchng article, so I crosschecked it against OP's CSV. The votes in the article do agree with OP's CSV (although the number of accredited voters differs slightly).
The crosschecked results are in KADUNA_crosschecked on line 3800. The image is here: https://inec-cvr-cache.s3.eu-west-1.amazonaws.com/cached/res...
Accredited voters: 276, Registered voters: 750, APC: 98, LP: 54, PDP: 102, NNPP: 11
All that said, I don't think that the results for 276 voters in one polling unit in one ward in one local government area in one state is clear evidence that Obi lacks support. If anything, the fact that OP's CSV matches a (potentially biased) news article gives me more faith in OP's tallies and claims.
(Aside: it seems _easier_ to lose an election in your own polling unit, where variance plays a larger part, than it is to lose on a wider scale.)