Can we add tests on non-seq generic objects in a generic proc and also normal object in a block expression.

Requiring mixin here might require every proc that would use reset to also have a mixin =destroy.
See:

• List comprehensions do not work with generic parameter #5707 list comprehensions requires mixin with generics
• Generics proc + macros: identifier resolution happens before macros #6387 Custom seq requires mixin (but iirc seq doesn't)
• mixin + block expression: works in Generics, not in "normal" proc #7385 Mixin in a block expression only works for generic object.

Meta-issue: #8677 (Generics early symbol resolution)

Requiring mixin here might require every proc that would use reset to also have a mixin =destroy.

That's not how mixin works.

In generics case it sometimes just does not work, here is another example i'm struggling with: status-im/nim-serialization#4

And the problematic type is a mixin in the template: https://github.com/status-im/nim-serialization/blob/6804ea25372de5919ef87cf13c7544770aacba2f/serialization.nim#L73

I don't think mixin comments are relevant . =destroy invocation for generic type needs a mixin. Period. There is nothing reset specific here.
I have extended the test to cover a type with no destructor.

What I'm saying is that if we have symbol visibility rules that requires mixin for this proc, we may have symbol visibility issues where the mixin within reset doesn't work and the package author will have to mixin =destroy in his own proc.