Clarification: floating point minimal representation

[leocencetti]

The section about type-to-format conversion in the spec states the following:

If an object can be represented in multiple possible output formats, serializers SHOULD use the format which represents the data in the smallest number of bytes.

The wording makes it unclear whether the set of "multiple possible output formats" is limited to the scope of the value type.

For example, let's assume we want to serialize the following C variable:

double value = 70.0;

Now, what should the serializer do?

1. Serialize the value as float32 instead of the full precision float64
2. Serialize the value as fixint since 70.0 == 70

According to the quoted extract, both approaches might seem valid.