no_std + async support

[ivmarkov]

As discussed in the matter-rs Matrix room, this is the sequential branch (i.e. no_std + async support).

This PR contains all changes which are already in the no_std and sequential branches of the matter-rs repository, plus the following additions:

• Built-in support for the embassy-net baremetal async networking library (under the embassy-net feature) - i.e. all necessary dependencies, plus a UdpListener implementation
• Removed dependency on smol. We only depend on async-io now, and only in the STD UdpListener. We can now also trivially implement a UdpListener for tokio's tokio-net as well, under a tokio feature flag
• Several simplifications which would hopefully make the examples a bit shorter and easier to understand. Namely:
  • Unification of the Matter, Transport and TransportRunner structs into just Matter, as was in the original main branch
  • Simplified Mdns implementation. MdnsRunner is gone. Just call MdnsRunner::run(...) to run the Mdns responder async loop
  • Simplified Psm implementation. Call Psm::new(&matter, dir) to initialize the persister, and then psm.run() to run the persister async loop
• UPDATE: Should be Apache 2.0 only - for now. IMPORTANT: Removed ./LICENSE and replaced with ./LICENSE-APACHE and LICENSE-MIT. @kedars I spotted this by accident. Basically, Cargo.toml used to say license=mit, while the LICENSE file used to contain the Apache 2.0 license. I assume you want both (as is the case for 99% of the Rust crates). If you want either Apache-only or MIT-only, please let me know, and I'll remove one or the other

[ivmarkov]

I've also switched the openssl dependency to the latest published one (and fixed both the mbedtls and openssl builds).

So upon rename of matter to rs-matter (and the git project itself) we can probably publish in crates.io

[ivmarkov]

@kedars Not sure from where the "restyled" thing is kicking in - couldn't see it in the GitHub Actions YAML. But in any case I think we just need to switch to cargo fmt --check instead.

[ivmarkov]

By the way, the changeset is so big because it contains also all the commits of the no_std branch. We can squash bits and pieces if you like.