More Optimizations

[torch2424]

From @MaxGraey in the Assemblyscript slack:

btw about opts. You could significantly speedup interop here and similar places:
https://github.com/torch2424/as-bind/blob/master/lib/supported-ref-types.js#L128

just use wasmExports.__getFloat32Array(responseRef).slice() instead
Float32Array.from(wasmExports.__getArray(responseRef))
Sometimes you could even don’t use slice if that result will use immediately) Adding .slice() (cloning) just for safety

similar for FLOAT64ARRAY, UINT32ARRAY and etc

[torch2424]

Oh this is actually a duplicate of #2 😂