unicode-width no longer depends on `std`. This has been updated inREADME.md in96eaa4a, but the introductory comments in `lib.rs` stillshow the old information.Align `lib.rs` with `README.md` and drop reference to the old `no_std`feature flag.

The package is always compiled as `no_std` and pulls in `std`  and`test` when compiling tests. Hence:- The prelude of `core` is available unconditionally, and it includes  `core::options::Option` and its variants. No need to manually declare  them.- The `test` crate is in-scope automatically, so no need to manually  declare it.

The `std` crate is unconditionally pulled in for test builds, hence wecan rely on it. Drop the redundant guards that check against `no_std`.

The `no_std` feature flag has no effect, anymore (as already documentedin `README.md`). Document this in `Cargo.toml`, but retain it forbackwards compatibility.Note that for better cross-package composability, an `std` flag wouldlikely be preferred in the future, over a `no_std` flag. The formerallows multiple packages with different preferences on this feature tobe combined in a single build, while the latter does not.Hence, suggest that `no_std` as feature flag is a legacy compatibilityflag, and will not be used in the future.