Specialize codegen for case of exactly one ink! message/constructor + that one being marked as wildcard

[cmichi]

Follow-up to #1020.

There is a possibility to optimize our codegen for this case:

• a contract containing exactly one ink! message (or constructor)
• that one ink! message (or constructor) being marked with the wildcard selector

In that case of our codegen currently still creates dispatch code for both the wildcard selector, as well as the auto-generated four byte selector. We should specialize the codegen for this case, this will enable us to drop parts of the decoding of the selector and input parameters. This will decrease the Wasm size for contracts that match this scenario.