Revert setting the default width of a serialized enumeration

[kheaactua]

Description of change

Setting the enum width as this code did narrows the enumeration value when it has a non-default width and a deployable object is unused.

e.g. That static_cast narrows the value of the enumeration regardless what its actual literal type is. This causes serialization to fail.

I first reported this in #36

Background

In #36 I show the fidl/fdepl used to define an enum with a non-default type, and show that the generated code (by generating with core, someip, then core again, as recommended in COVESA/capicxx-core-tools#32, versions 3.2.14) has the proper enum literal type. However that line causes the deserialization to fail.

This is due to this change in OutputStream.hpp

-   writeValue(_value.value_, static_cast<EmptyDeployment *>(nullptr)); 
+   // Default enumeration width is 1 Byte
+   uint8_t value = static_cast<uint8_t>(_value);
+   writeValue(value, static_cast<EmptyDeployment *>(nullptr)); 

and was committed with the comment:

Changed the default width of a serialized enumeration from "backing type width" to 1 byte.

Impact

With this change, my deployment type is still a uint8_t, e.g. (from #36)

typedef CommonAPI::SomeIP::EnumerationDeployment<uint8_t> ColorDeployment_t;

And the Deployable is still set to nullptr on creation, which is expected behaviour according to COVESA/capicxx-someip-tools#37, though it means we skip the entire code path that deals with a deployable.

But OutputStream::writeValue(const Enumeration<Base_> &_value) is able to pass the proper value (as type information is retained) to writeValue, which fixes serialization.

[kheaactua]

I'm confused why there is so much code guarded with if (_depl != nullptr), but as far as I can tell there is no way for _depl to be anything but nullptr. Am I missing something? I haven't found anything in the examples, CommonAPI docs, or Franca docs yet, but I'm still looking.