[core] Remove Floating UI dependency?

[oliviertassinari]

Summary

I expect many developers will ask themselves, why do we need to pay for Floating UI bundle size when the lowest version of the browser we support has the CSS "anchor position" feature available (not true yet): https://developer.chrome.com/blog/new-in-web-ui-io-2024#anchor_positioning.

So I think that having this issue to work toward removing the logic not needed anymore will at worst bring clarity to developers, and at best reduce the library size.

If there are still strong reasons to have Floating UI, then I think:

• We could have a mode with it, allowing developers to optionally pay the cost
• Report those limitations to the W3C, they might want to fix them

Examples

• Sholace exploring new browser APIs to simplify tooltip https://x.com/claviska/status/1790010420494807288 - https://github.com/shoelace-style/shoelace/blob/77d7ce5f2fc87121c7f7b0e5e3e09fbdc48f587a/src/components/popup/popup.component.ts#L1
• CSS Anchoring API 👋 floating-ui/floating-ui#2074 discuss overall limitations for Floating UI, but I imagine we are using a subset of those in Base UI.

Motivation

I'm going through https://developer.chrome.com/blog/new-in-web-ui-io-2024, and reporting all the opportunities that have no issues, so we keep the library at the edge of what's technically possible with today's browsers.

Search keywords: -

[mj12albert]

One of the main points in #1246 is to reduce the duplication between Base UI and Floating UI, should we keep this one open? @atomiks @michaldudak

[mj12albert]

Also related: #454

[mj12albert]

We discussed this and decided to close this issue as it will likely be 2-3 years or even more before native CSS has parity with the Floating UI features we are using, so there's no need to track this for now