Derive Arbitrary impls for a bunch of chain and network types

[teor2345]

Motivation

We need Arbitrary impls to write proptests for internal and external network protocol messages.

Solution

• Derive Arbitrary impls for a bunch of chain and network types
• Add proptest Strategy impls for u32 and full datetimes

The code in this pull request has:

• Documentation Comments
• Enables Unit Tests and Property Tests

Review

Anyone can review, as long as they understand proptests.

Related Issues

Split off #2160

These tests might have caught security issue #2148

Follow Up Work

Rest of #2160

[teor2345]

I think the macOS sync_large_checkpoints_mainnet failure was caused by #2172 or #2154, and it should be fixed by #2181 or #2182.

I don't think it's related to this PR, because this PR only changes proptests.

https://github.com/ZcashFoundation/zebra/pull/2179/checks?check_run_id=2635749551#step:10:644

[dconnolly]

The zebra-chain::primitives module has been a place for cryptographic primitives such as the signature schemes and the proof types, rather than 'utils' or other base types. Sometime like Amount got its own module. Should we have a top-level arbitrary module for things like this?

[dconnolly]

@teor2345 The zebra-chain::primitives module has been a place for cryptographic primitives such as the signature schemes and the proof types, rather than 'utils' or other base types. Sometime like Amount got its own module. Should we have a top-level arbitrary module for things like this?

[teor2345]

@teor2345 The zebra-chain::primitives module has been a place for cryptographic primitives such as the signature schemes and the proof types, rather than 'utils' or other base types. Sometime like Amount got its own module. Should we have a top-level arbitrary module for things like this?

These particular strategies are serialization-specific and consensus-critical, so I moved them to serialization::arbitrary.

[teor2345]

@jvff this should be ready to merge if you want to use it in the #2178 tests