Release new version 1.6.3 on PyPI

[bittner]

A stable version of Autograd was last released on PyPI on 23 June 2023 (commit 1bb5cbc). Afterwards, a few investments were made that make sure Autograd can continue to be installed successfully (e.g. #602). Those changes are still unreleased.

It would help if a new release on PyPI were made.

---

P.S.: As explained in #605 (comment), the only thing that needs to be done is push a Git tag in order to trigger the automatic release process. As a one-time prerequisite, credentials from PyPI need to be configured on GitHub beforehand.

[agriyakhetarpal]

An environment for PyPI has now been set up. @fjosw and I shall be looking into getting this done in the coming days. We were thinking of incorporating NumPy v2 support (#618) with it, so we might as well bump it to version 1.7.0. @j-towns, I question I have is whether it would be a good idea to do a pre-release first with NumPy v2.0.0rc2 or if we should bump to v2.0.0 stable directly? I suppose the latter would be better since ABI remains stable between both, and I've found that it's easier to work with bounds when they don't specify pre-releases.

[j-towns]

Bumping straight to 2.0.0 sounds simpler, so I would probably do that.

[agriyakhetarpal]

New release is up, though we bumped to v1.7.0 directly: https://github.com/HIPS/autograd/releases/tag/v1.7.0 🎉 so we can close this.