Consider a single default convention

I am genuinely interested in learning about the use cases of custom conventions. Do you have practical use cases where users needed to customize the default convention? What is the value of having multiple conventions? Does it actually help the community converge into something? My impression is that everyone is using DefaultConvention and we are maintaining code that is unnecessarily complex with support for multiple conventions. If a package A relies on DefaultConvention and package B relies on CustomConvention, what happens? Do we really want to support such use cases?

Appreciate if you can clarify these questions. I can help with https://github.com/JuliaAI/ScientificTypesBase.jl/issues/21 after this is sorted out. My personal opinion is that we should not waste time modeling multiple conventions and should ask the community to adhere to a single convention for scientific types. It has tremendous benefits in more complex pipelines.

No file, test, or implementation entry point is named; start by reading the issue discussion about DefaultConvention and CustomConvention. Review ScientificTypesBase.jl issue 21 only after the convention question is resolved. Done means reaching a decision on whether multiple conventions should remain supported, with the resulting scope made explicit.