Use error enums instead of Anyhow

[algesten]

Original discussion on discord, and further issue raised here: webrtc-rs/rtp#14

Anyhow is great, but maybe not a good fit for webrtc.rs. To summarize:

• Anyhow is mostly meant for applications, not libraries.
• Anyhow doesn't implement std::error::Error (it can't), which makes it hard to interoperate in apps that rely on that.
• Not having explicit enumerations of possible errors means the webrtc.rs API effectively have "hidden" code paths. It's not possible for a user to know which errors could potentially be thrown from an API call returning anyhow::Result<X>

Way forward

• Replace anyhow with idiomatic Rust error enums.
• Keep thiserror to help implement said enums.
• Maintain ergonomics for ? use internally in webrtc.rs (work with From traits and rewrap errors).
• Bring out possible errors in API calls. Either via types or documentation.

This issue will be used to coordinate the effort which will span all the sub-crates.

• sdp
• util (done, but deliberately marked as draft)
• mdns
• stun
• turn
• ice
• dtls
• rtp
• rtcp
• srtp
• sctp
• media
• interceptor
• data
• webrtc

[algesten]

First stab at sdp is done. Next up I'll do util. I want to do from the "bottom up" following the dependency tree, and we won't get a good idea of how this PR will work before we get to the second/third level of dependencies.

[algesten]

@rainliu Would you mind looking at the two first PRs? I'm wondering what you think about it so far. I've hard to maintain the convenience of ? and PartialEq.

[rainliu]

@algesten, thanks for the efforts.

PR for sdp crate looks good to me.

and I leave some comments on PR for util crate.

[algesten]

Alright. That's all for me for today. I'll continue tomorrow.