A truncate operation involves more context than many other operations, and is a bit unusual in that it is a data-modifying API on the WT_SESSION handle, rather than on a WT_CURSOR.

For that reason the truncate implementation from an API layering perspective is a bit unusual. It can be complex to follow the logic about the API, and that's made more difficult because different parts of the truncate implementation take different sets of parameters.

I propose that we create a new WT_TRUNCATE_INFO data structure to store context about the truncate operation. This was prompted by work on WT-9715