It was noticed that the code responsible for merging in defaults in the operator does not work as expected in all situations (to be more precise: as it is currently used it doesn't differentiate between "value not specified" (in a user config) and "value is a pointer to an empty value" (think e.g.: a pointer to a boolean false). This is problematic, because essentially means that a user-provided explicit "false" will be treated as "not specified" and hence potentially overwritten with whatever the configured default is.

Acceptance Criteria:

The code merging in defaults must be fixed so that it does not overwrite "pointers to empty (leaf!) values" with defaults. Note that a pointer to an empty sub struct must not prevent traversing into the nested struct and potentially filling in defaults there.