[future] consider dropping send/sync/static trait bounds once possible

[GlenDC]

There might be a future where:

1. the trait resolver and friends become better in detecting such trait bounds and we do not need to declare them everywhere
2. we no longer aim to support tokio multi-threaded work-stealing but instead have structured concurrrency (e.g. 1 runtime per thread0
3. we become better at coding (or a contributor appears on the stage who is) and finds a more ergonomic way to write this

It would allow us to write simpler code, where we also can work more often with references (e.g. not tied to 'static) instead of having to either resort to non-sync or use the pass-the-ball pattern (where you return the args back via result... dirty).

It's a word I do want to live in, but at the time it was too much of a distraction and either way Rust async still has plenty of work. This is more of a zombie tracker so we can look into this somewhere mid/late 2025. And maybe it just turns out to be a "grass is greener on the other side" kind of issue that we drop as won't do. We'll see.

Also feel free to start discussions here related to these bounds and the use in rama.