Cleanup address deserialization

[lehins]

Description

Currently there are 4 separate implementations for deserializing an Addr. I am not gonna go into history why this is the case, but this PR moves old implementations into tests and uses them for ensuring that deserialization doesn't break accidentally over time and also as a baseline for checking performance. Summary:

• There is now one single deserialization implementation for both CompactAddr and Addr
• Constructor for CompactAddr is no longer exposed, making it impossible to create invalid and dangerous binary representations for addresses
• Combines Cardano.Ledger.CompactAddress into Cardano.Ledger.Address, while preserving git history
• There is no longer a need to supply a version for rountrip tests, they will iterate through all versions of serialization, since that property must hold for all versions
• Switch to fromCBOR in BabbageTxBody and BabbagetxOut for address. Now possible thanks to versioned deserialization

Checklist

• Commit sequence broadly makes sense
• Commits have useful messages
• New tests are added if needed and existing tests are updated
• Any changes are noted in the changelog
• Code is formatted with ormolu (which can be run with scripts/ormolise.sh
• Self-reviewed the diff

[teodanciu]

Looks good to me.
First two commits could be probably squashed into one, to make each remaining commit work correctly, and same with next two commits, but don't think it matters too much, with all the moving around.

[lehins]

First two commits could be probably squashed into one, to make each remaining commit work correctly, and same with next two commits

I was unable to squash last two commits, since we then loose git history, however squashing the subsequent two worked out fine.