Add support for NumPy data types other than `float32` to `DynamicVamana.build` in Python

Currently, in Python, `DynamicVamana.build` always creates an index with `float32` data, even when given an array with other types, such as `float16` or `int8`. We should add support for the other [NumPy types `Vamana` supports for building](https://intel.github.io/ScalableVectorSearch/python/vamana.html#svs.Vamana.build). Compressed non-NumPy types can be considered in future issues.