Fix numpy version and remove scipy optional

[tylerhutcherson]

Original goal of the task was to relax the forced numpy version constraint that was causing issues in Python 3.11 environments where numpy 2.x.x was required (like Google Colab)

Problem

However, I quickly ran into dependency chain issues with:

• scipy -- the scientific computing stack is a bit like a dinosaur still
• boto3 -- always an issue
• ranx -- not really sure why, but this one has 13+ separate dependencies and includes a pin on numba that expects certain things from numpy and then it gets easily circular

Solution:

Working together with claude, I was able to resolve and rebuild the environment and the lockfile with one major change.

• ranx is still an optional, but it's not an explicit extra.
• What does that mean? Literally only means that if devs are using ranx they have to install it with pip install ranx instead of pip install redisvl[ranx]

This could be some kind of limitation with the poetry version I was using. But I spent 6 hours on this with Claude and probably spent 50$ in credits on working through an onion of python environment. It now works and our issues on google colab are gone. I also think the package ranges are a bit more generous now.

[Copilot]

Pull Request Overview

This PR streamlines dependency declarations by consolidating the NumPy version constraint and removing SciPy as an optional dependency, while updating related extras.

• Consolidated NumPy requirement to a single version specifier.
• Removed conditional SciPy entries and cleaned up extras.
• Corrected hyphenation for the google-cloud-aiplatform extra.

[abrookins]