I would suggest that the passenger(s) must already be intoxicated or under the influence prior to boarding to satisfy a violation under 91.17(b). There was a good article about a 91.17(b) violation in the latest edition of AOPA Pilot, where the violation was tacked on as another charge to a different violation, but the NTSB trial judge dropped the violation.
If you look at the article again (the case itself has not appeared on the NTSB opinion web site), I think you are probably right about the 91.17(b) charge being "tacked on." Looks like the primary charges involved the FAA claiming it was a gray charter.
But if you think the 91.17(b) charge went away, you'd be wrong. Read the article again. The ALJ dismissed all of the charges, apparently saying that 91.17(b) did not apply to a pure Part 91 flight. The FAA appealed and the NTSB panel decided that the ALJ was wrong and 91.17(b) =did= apply.
As to the passenger that =becomes= intoxicated during flight, your guess is as good as mine, but the language of the reg
==============================
Except in an emergency, no pilot of a civil aircraft may allow a person who appears to be intoxicated or who demonstrates by manner or physical indications that the individual is under the influence of drugs (except a medical patient under proper care) to be carried in that aircraft.
==============================
ends with "to be carried in" not "to board" making it susceptible to an interpretation that it applies to someone who gets drunk on board.
The problem with this is that regs like these that have not been interpreted before tend to get interpreted when a problem comes up - like an inebriated passenger causing problems - and the FAA's interpretation, so long as not ridiculous, is the one that counts.