You say that plural abbreviations (e.g. paras) require no full stop because the last letter is the same, which is quite logical. This implies we get e.g. "no. 1 and nos 2&3" and "fig. 1 (UK consumption of figs) and figs 2&3 (UK/EU consumption of dates)", right? You see the potential for confusion, I'm sure. Yet nos./figs. and no.s/fig.s both look wrong, so I assume there's no choice. I'd be interested in your view (particularly if there's any corroboration from other sources)...
I'm afraid, you've caught me out. When writing Grammar Monster, I noticed
that many of the leading grammar references contradicted each other in a
number of areas. Furthermore, over the years, I have learnt that as you
(apologies for using 'you', but I am on a personal crusade to have 'one'
removed from the English language) become more involved with grammar, you
can justify what many consider blatant errors, just through being aware of
the cases when exceptions are feasible.
With regard to your point, I found this "law" in an old grammar book from
the 40s, which I discovered at a car-boot sale. (It's at my other house, so
I can't give further details - sorry.) I loved the rule. As no other grammar
references gave clear guidance, I went firm on the ruling for Grammar
Monster.
The purpose of Grammar Monster is to cover most of the points for those
required to produce grammatically credible correspondence. In order to make
it readable, I had to summarise many rules into true but "palatable"
versions. The nos./nos question does not really fall into that category, but
you can see that Grammar Monster was never supposed to agree with the
leading grammar references. After all, they can't agree. I suppose, I just
liked the rule. Moreover, I suspect it was the rule once, but has changed
since. I want it to be the rule, I guess.