(1) I'd take a digital image of the photograph, and then write a citation for it and include that citation on the digital image. That citation would report the photograph was in the grandfather's album,; would improve supplemental comments that M or MS provided about the both the album and the photograph, including that the identification was made by M and MS (I'd say when and where).
(2) I'd enter "Aunt Bertha" as an unrelated individual to my family file, citing the photograph, citation I'd written earlier. Other than the name "Bertha," I'd probably have only one tag, and that tag would be about the photograph. I'd link that same tag to my grandfather.
(3) When I found the second photograph (gravestones) in my "papers," I'd take a digital image of it and write a citation for it. Using that second citation, I'd use one of my "who's this" research tags and enter a note about same to that unrelated "Aunt Bertha" person in my file.
(4) I'd NOT "infer" any relationship between "Bertha" and the others. (If there was a relationship, cemetery records or obituaries would confirm that. For me, information in a research note would be enough.)
(5) When I learned the original picture of AB was later than originally thought, I'd update my citations (digital image and database); ditto the date on the photo tag. Depending on the notes I'd made about the gravestone image (who's this tag), I'd update that information also.
I work from the known to the unknown. Bertha would still qualify as an unknown to me. I'd invest my time learning more and more about my grandfathers family, believing that research would ultimately lead me to a more complete record of his relations.
I'll try and recall what I did in my software in that case:
(1)
I made a new person for Aunt Bertha, with the infos that I had.
I added a husband with the surname J.
I added the daughter and her relationship as a godparent to MS.
(3)
I went to Elisabeth N, née K (she and her husband Wilhelm were already in the database).
I added the daugther Luise and her husband Friedrich.
I added their daughter Bertha H.
(3,5)
I nearly decided to merge both Berthas - for a while I thought it was a "slam dunk" (ouch. I must have been tired.)
Luckily, I just wrote Notes for both Berthas where I put down my idea that these might be one and the same.
(4)
Now I was happy I still had both persons seperate. I rewrote the notes to remind me that these two most probably were not the same person.
I also added a nickname "H Aunt" to Bertha H.
(5)
I changed my notes again to show the "disproof-statement" for my idea...
Until I find the real connection of AB to our family, I will leave the notes in both Berthas. So that in the future, nobody makes the same mistake as I did, not even me, if I forget this story.