Consider reverting `length` safety improvement about extism

[G4Vi]

That sounds good. Let's some to some consensus with the team on naming of ExtismHandle before we proceed.

[zshipko]

I'm not opposed to reverting that change, we should just make sure to add a note somewhere that it should only be called on ExtismHandle values.

[zshipko]

We discussed making this change today, but when I was in the process of testing things out I realized there are some places where we make the assumption that non-handle offsets will return 0 for extism_length, such as https://github.com/extism/rust-pdk/blob/44370c59677037f71d4911ecd65de0af7e0e3de0/src/memory.rs#L42

This is only really an issue when taking input into a host function, where it's not clear if the offset was created using extism_alloc or not. I think it's best to keep the slightly more complex extism_length implementation for now. I will give this a little more thought this week.