chore: preserve original resolve when decoding WIT package twice

[0xsenty]

The second call to wit_component::decode reused the resolve identifier from the first decode. This silently overwrites the original resolve, which can be misleading — especially since both resolve and resolve2 are referenced later in the function.

I’ve updated the second match arm to bind to resolve2 instead, so both decoded values are preserved independently as originally intended.

This should clarify the logic and avoid unintentional data loss.

[alexcrichton]

Thanks! Is this perhaps the wrong change though? This current PR looks like it's a small refactoring but no actual behavior changes.