I don’t think you achieve what you’re looking for in a single name record.
A “name” is highly culture-related and its definition varies across space and time.
A “name” in Gramps is based on several “components”:
- inherited: will be stored in surname (may be family name, patrilineal, patronymic, matri-…, …)
- intentionally received: usually given names
- chosen deliberately by the individual: a pseudonym
- attributed by others: nickname
- honorific: this is the case in aristocracy with name of possession
- …
Gramps tries to fold these “components” into record fields to allow some processing with the bits. A Gramps “name” is supposed to be a “closed form” to designate a person. Given names are all grouped together as a single logical component. Obviously this does not fit with your goal where you have variants coming from different sources. Don’t try to twist name fields into your schema. You’ll most likely create mess where names are already something complicated.
The surname can be split into individual “atoms” but it is mainly intended for specific rules like:
- Spanish usage where children get father and mother family names, mother name being dropped on next generation
- aristocracy: notably German usage where the possession are preceded by “von” or “zu” depending on the kind of authority they have and linked by “und”
In such a case (multiple conflicting names), I record several names because if he is designated as Andra, it won’t simultaneously be called Chish. I’d tag Petro Anotonv Andra as the preferred name and have other entries for Petro Nikolaev Andra and Petro Antonov Chish.
Having separate name records allows to associate them with specific citations. And citations have a confidence level which could help with your research.