When should FetchClient be preferred over BrowserClient?

[EchoEllet]

The default http package implementation already uses fetch() on the web as of 1.3.0 (PR dart-lang/http#1401).
Previously, it used XHR.

I appreciate the effort behind maintaining this library. Since BrowserClient now uses fetch(), when should FetchClient be preferred?

It appears to expose additional Fetch API features and configuration options.

[Zekfad]

Initially this project started as a way to allow reading response as a true stream without waiting for XHR to collect full blob. And it was a nice journey for me personally, as I watched JS and Wasm bindings mature over the lifespan of a single library.

As of today this library:

• Still works like a normal client.
• Adds platform specific configurations, such as allowing or restricting CORS.
• Adds a probe mode to approximate redirects.
• Allows to opt in into request streaming for supported browsers and servers.
• Exposes Fetch API related enumerations for VM platforms (easier to customize client via kIsWeb or kIsWasm, without creating conditional import and stub files).

It should also work in NodeJS-like environments where fetch is available.

So this boils down to platform specific options.

As for updating readme, I'm not so sure if this is needed, since all I said is already there, but I'm open for suggestions. And if I'll be updating the package, it worth to add http_profile integration.