Cardano Updates

Fix nightly CI

Somehow whenever `StakePoolsSnapShot` is created there is a thunk that
gets into it. I can't quite figure out what is causing it, but
concidering it only happens during generation of arbitrary data and for
the initial test state this shouls be fine foce 'em. It would be nice to
get to the bottom of this, since theoretically thunks should not be possible
in `VMap`. Potentially switching to using a strict tuple could help.

That field being lazy could not lead to space leaks, but it was forced
upon construction of `SnapShot` anyways, since another strict field
`spssStakeRatio` used this value upon construction.
Therefore, for simplicity and consistency sake it makes sense to keep
this field strict.

Previous implementation was wrongfully iterating just over the vector of
values, which itself is lazy, causing unexpected thunk buildup.
Technically it is a nice optimization, since that would mean there is no
iteration over the vector of keys, however, without forcing everyone to
switch to newest version of `vector-0.13.2.0` that introduced
`Data.Vector.Strict` we cannot have a strict `map` (using `mapM` is out
of the question, since it has terrible perfomrance)

This commit also adds strictness tests for `map` and `mapWithKey` and a
`noThunks` instance for future compatibility

Add `Generic` instances for `ApplyTxError` type family

* Remove WIP from Conway spec; add explanation of formal specs

* Apply suggestions from code review

---------

Co-authored-by: teodanciu <[email protected]>

Co-authored-by: teodanciu <[email protected]>

New namespace, that follows the same patterns as utxo and blocks
namespace.

As it was possible to abstract code and make it depend only on era
it was possible to remove era specific code. After this change
'cardano-ledger-canonical-state:conway' may only keep the instaces
for the era spefic types.

Co-authored-by: Alexey Kuleshevich <[email protected]>