feat: require chain_id be 32 bytes

[SuperFluffy]

Summary

Make chain IDs 32 bytes and construct them from sha256-hashed names.

Background

At the moment chain IDs sent as part of sequence actions can be arbitrarily long. The only place where an upper limit on chain IDs is enforced is in SequencerBlockData, which however is constructed in sequencer relayer and after cometbft/sequencer have come to a consensus over new data.

Using a chain ID of 32 bytes as opposed to a vec is desirable for the these reasons:

• it allows it to be Copy so that unnecessary vec-allocations are avoided
• it minimizes the likelihood of rollup namespace clashes because it suggests to use sha256 (although it cannot prevent it)
• it allows to deterministically calculate the overhead for each transaction allowing to take it into account when calculating tx submission fees

Changes

• Make ChainId a think wrapper over [u8; 32] (away from Vec<u8>)
• Move ChainId to astria-proto
• Reimplement all services in terms of it (sequencer now enforces the length through ChainId being a part of SequenceAction, whereas before it was just a byte vector)
• Adjust all tests

Testing

All tests still pass or have been adjusted. The snapshot test for calculating the merkle tree hash over the actions had to be updated.

Breaking Changelist

This breaks all public APIs because chain IDs are now expected to be 32 bytes (whereas they were arbitrary length before).

[noot]

looks good to me!