Custom multiplier is performing type casting before executing the step.
This causes the following results:
In this example, received value is 0.9, custom multiplier is set to 100 and result should be stored as Numeric (unsigned).
Expected result is 90, while current result is 0.
This is caused by the fact that custom multiplier casts received value to the type of information of the item (Numeric (unsigned)), at this point 0.9 becomes 0.
Limitations comes from the fact that we do need to support both floats and unsigned integers (different precision and max value). I propose to at least document such behaviour.