Calls from different radio amateurs using the same equipment are fine to count separately. After all, be they husband/wife or father/son, they are most probably both SOTA chasers wanting the contact each for themselves. So that cannot be a problem.
Same operator, same equipment, different callsign (eg club call or previous 2E0 call for example). Well, I would say that each distinct call represents a different station, so it should be allowed. It isn’t ideal, or indeed very satisfying, but at the end of the day, the decision can only rest with the activator, unless we are to enter an enormous database linking different licence class and club callsigns within the current database!
It is rarely an issue for me these days, with increasing interest in working SOTA stations, and my preference for HF CW. But for many years I tried to activate everything on 2m FM, and away from population centres, on weekdays, 3 or more years ago, this could be difficult.
On Shobdon G/WB-017 in 2006, Jimmy and I were stuck on 3 contacts. One of the strong locals we had already worked came back on and offered up his other callsign. It was 7pm at the end of a long multi-summit day, and I decided it was better to accept the summit qualification and get some food, rather than hang around calling for another.
On the Pennine Way in the same year, Jimmy and I were struggling on Great Shunner Fell G/NP-006. The weather was appalling, and so a handheld activation was in order. IIRC, our 4 contacts were with a station, the same chap using the club callsign registered to him, another station, and his licensed daughter! We were on the verge of abandoning the activation as unqualified anyway, so it was nice to least get the four QSOs, by hook or by crook!
To the original issue, at one end of the amateur radio contact, you can only take things in good faith. Unless you have good reason to question it anyway. I was called by a “ZL” on The Cloud last year. I was most suspicious and didn’t log it, but made a note. All investigations suggested that the call was false.