Cardano Updates

* Relax `NoThunks` instance for `BlockTransitionError`:

We no longer need to check for thunks in the contents of
`BlockTransitionError`, becasue the constructor uses `deepseq` to force
everything to normal form.

We could of course bring back strictness annotation and `NoThunks`
instances for all of the predicate failures, however this whole overhead
is an overkill, because:
* we do not need to optimize for the failing case and it is ok to
  introduce a little overhead upon failure. FTR: the only time we
  persist predicate failures is when we write to disk temporarely for
  chains with bad blocks.
* Maintaining `NoThunks` instances for *all* predicate failures for a
  single edge is a burden
* There is a plan to utilize laziness in predicate failures in order to
  add timing measurement for failure checks in ledger rules

We no longer need to check for thunks in the contents of
`BlockTransitionError`, becasue the constructor uses `deepseq` to force
everything to normal form.

We could of course bring back strictness annotation and `NoThunks`
instances for all of the predicate failures, however this whole overhead
is an overkill, because:
* we do not need to optimize for the failing case and it is ok to
  introduce a little overhead upon failure. FTR: the only time we
  persist predicate failures is when we write to disk temporarely for
  chains with bad blocks.
* Maintaining `NoThunks` instances for *all* predicate failures for a
  single edge is a burden
* There is a plan to utilize laziness in predicate failures in order to
  add timing measurement for failure checks in ledger rules

Co-authored-by: Fabrizio Ferrai <[email protected]>

This will use the correct package database

It doesn't work reliably and is incompatible with a nix environment

This brings in https://github.com/input-output-hk/haskell.nix/pull/2498
which is needed for doctesting in nix

Add golden tests for the rest of the query result types

This commit introduces a general hierarchy for common era testing

Fail PlutusV1-V3 translation for non-empty `subTransactionsTxBody`

Triggered by:

```
cabal run cardano-ledger-conway:test:tests -- --match "/Conway era-specific/CDDL/Huddle/Decoding fails when term is zapped/plutus_data: MemoBytes (PlutusData ConwayEra)/" --seed 1615161393
```

The issue is that our list of allowed tags for the `Constr` data
constructor of Plutus `Data` is not exhaustive enough. As per
`PlutusCore.Data`, the allowed tags are 121-127, 1280-1400, and 102.