Consider improving Tiny Solver in upstream Ceres instead of making a fork

[keir]

Hi tiny_nnls_solver maintainers,

I'm the original author of Tiny Solver in Ceres. I'm happy to see you've made some improvements to it! We would be more than happy to work with you to improve the upstream version of Tiny Solver instead of having this fork. There are several benefits to being upstream:

• Upstream code review - This is good or bad depending on your perspective, but we have very strict requirements around code in upstream. This can help improve the quality of what you've started by having us participate in your improvements.
• Bigger reach - Ceres is well known at this point; by being part of upstream, the probability increases that your code will create impact beyond your own project.
• Math rendering in docs - Our docs have proper math rendering; so we could have a nice separate doc for Tiny Solver. There currently isn't anything specific to Tiny Solver in the docs, but only since no one has done it yet.
• Continuous test integration - Ceres gets extensively tested as part of Google's CI suite, among other places.

We understand that upstream is not for everyone, and sometimes the code review process can be slow, but we've found that for foundational libraries like Ceres the slow and thorough approach is the right one. For this project, going upstream would mean creating a proposal for what parts to go upstream, then creating a slow trickle of patches to add the functionality in this fork upstream.

Cheers,
Keir