Refactor DmaBuf creation with explicit modifiers
Currently all DmaBufs created in Kwin are allocated using the old GBM api without explicit modifiers.
Aside from better compatibility when sharing buffer between different GPUs when screencasting is present there are also reports that the nvidia driver only supports allocation with explicit modifiers.
To use DmaBufs with explicit modifiers this property has to be plumbed through the whole stack.
Edited by Merlin Lex