Cleanup pointer serialization

[lehins]

Description

Previously definition of a pointer used Word64 type for slot number, transaction and certificate indices. We can finally use narrower types and treat bogus pointers as invalid for all eras retroactively. There were a few stages to this that were necessary over the last few years:

1. In Babbage era deserialization of transactions prevented pointers with values outside of Word32 for slot number and Word16 for transaction and certificate indices
2. Upon entry to Conway when translating ledger state all pointers that didn't fit in those bounded types without overflowing were converted to an all zero pointer: Ptr 0 0 0
3. Once in Conway (i.e. finally we are) we can pretend like all bogus pointers where always clamped to zero. The reason why we can do that retroactively because all those bogus pointers that exist on mainnet never affected reward calculation or anything else for that matter, so clamping them to zero permanently will not affect that in any way. In order to do that we need to do pointer normalization (i.e. clamping to zero of bogus pointers) during deserialization of transactions, instead of during translation into Conway.

Step 3 is implemented in this PR.

This was a long awaited cleanup that is finally here.

Checklist

• Commits in meaningful sequence and with useful messages
• Tests added or updated when needed
• CHANGELOG.md files updated for packages with externally visible changes
  
  New section is never added with the code changes. (See RELEASING.md)
• Versions updated in .cabal and CHANGELOG.md files when necessary, according to the
  versioning process.
• Version bounds in .cabal files updated when necessary
  
  If you change the bounds in a cabal file, that package itself must have a version increase. (See RELEASING.md)
• Code formatted (use scripts/fourmolize.sh)
• Cabal files formatted (use scripts/cabal-format.sh)
• hie.yaml updated (use scripts/gen-hie.sh)
• Self-reviewed the diff