Improve performance of `global.{set,get}` with externref

First implemented in https://github.com/bytecodealliance/wasmtime/pull/1969 the easiest thing at the time was calling out to intrinsic functions to implement `global.{get,set}`, which means that there are no JIT fast paths for these operations. At the time of this writing there's no toolchain that uses `global.{get,set}` so we don't have a ton of motivation to make this fast, but I wanted to file this issue to make sure we don't forget about it and in case it comes up in the future!

For now it's likely safe to not worry about the performance here, but if this becomes a hot spot in the future this can be a point of coordination!