Build_disk_index from Iterator

[J0SZ]

The documentation of build_disk_index states:

This function will construct a DiskANN disk index. Disk indices are ideal for very large datasets that are too large to fit in memory.

However, it only accepts

1. Numpy array: Which cannot by definition fit into memory
2. Path to vector bin file: In order to create this file, I once again need Numpy array: 1)

In other words, this function defeats its own purpose since in order to use it, I need to put all my data that are not supposed to fit into memory... into memory...

I would suggest to make it accept an Iterator[np.Array] or correct the documentation.

[hhy3]

You can write an iterator of np.array to a file sequentially