Document v1/v2 split and motivation for moving away from serde

[mwcampbell]

As a newcomer to this project, I was confused at first, because the README only barely mentioned the split between v1 and v2. Also, as far as I can tell, the motivation for moving away from serde in v2 isn't documented. I'm curious about the motivation because serde is a critical part of the Rust ecosystem, and I'm afraid that requiring structs to implement or derive bincode-specific traits will be a step backward for bincode, but I'm sure there's a good reason.

[BarbossHack]

note that you can still add the "serde" feature in v2

[VictorKoenders]

We have a list of known issues with serde on docs.rs. There are some other reasons we wanted to make the serde dependency optional (supporting embedded targets, reducing dependencies, increasing performance). We can add a FAQ to the repo with these reasons.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[richardpringle]

I'd love to see a little write-up into the decision-making process beyond solving for the "known issues" with serve in the docs.

This would help me for two reasons:

1. Help me decide whether bincode's goals as an encoding format are aligned with what I need it for
2. Help me re-evaluate some of my own decisions around the use of serde

It wouldn't need to be perfect nor even accurately reflect the discussions that took place, but it there's something that can help with the above (namely 2, I think most of your other documentation covers 1, but it would still give me some insight), that would be appreciated.