This is to use zero-copy deserialization from protobuf encoded bytes to the generated Rust types, and to make conversion between the generated Rust types and idiomatic Astria types cheaper.

This issue has a few steps:

1. Update the protobuf compilation tool to emit Bytes instead of Vec<u8>
2. (generate new code by running the tool)
3. Update all conversions in astria-core between the protobuf and native domain types (primarily all try_from_raw, try_from_raw_ref, into_raw, and to_raw).
4. Update all code that broke because it required Vec<u8> (I expect this primarily affecting tests).

For step 3 in particular care should be taken to not end up with conversions Bytes -> Vec<u8> -> Bytes because we have some instances where we convert Bytes to Vec<u8>.