Restructure the Core for consistency, clarity and new functionality

[lehins]

Changes in this humongous PR:

Creation of class hierarchy:

New classes:

EraSegWits --> EraTx --> EraTxBody --> EraTxOut --> Era
                     \              `--> EraPParams --> Era
                       `--> EraWitnesses --> EraScript --> Era
                        `--> EraAuxiliaryData --> Era

Note that all of the Era* classes have a few properties:

• depend either directly or transitively on the Era type class.
• define an associated type family (which was previously standalone).
• add all constraints that the associated type must implement.

This means that any function that uses an associated type family can safely use
the corresponding type class. For example a function that accepts TxBody as
argument can always set EraTx as a constraint, which will bring all of the
superclasses as constraints as well.

Loose corespondance of new classes to the old removed concepts:

• Era class is the same as before with the addition of type level protocol
  version and deprecation of all of the functions
• EraTxOut has all superclasses as in UsesTxOut type synonym constraint as
  well as defines lenses that supercede all functions previously defined in
  Era type class
• EraTxBody has all superclasses as in UsesTxBody type synonym constraint
• EraAuxiliaryData has all superclasses as in UsesAuxiliaryData type synonym
  constraint and consolidates ValidateAuxiliaryData into it
• EraPParams replaces UsesPParams type class now with an associated type family
• EraScript has replaced UsesScript type synonym constraint and the
  ValidateScripts type class, with an exception of validateScript function,
  which was moved into EraTx type class, since it depends on the Tx

Lenses

This PR expands usage of
microlens package.

Each class defines lenses for associated types, all of which are getters and
potentially setter. They also define a way to construct a minimal value of an
asscoiated type. For example if we want to construct a function that will
construct a Tx with a new TxBody, which uses AuxiliaryData from another
Tx:

newTxWithOtherAuxData :: EraTx era => TxBody era -> Tx era -> Tx era
newTxWithOtherAuxData txBody tx =
  mkBasicTx txBody & auxDataTxL .~ (tx ^. auxDataTxL)

Above function is guaranteed to work with transactions of all eras.

Some are just getters, which simply relay some information from one or more
fields of a concrete type, altohugh they might do some minimal computation. For
example allInputsTxBodyF will return all of the inputs present in the
TxBody.

Naming convention

All lenses are suffixed with the type name they operate on and either Lens or Getter.

Migration guide

Accessor fields

Names suffered some very minor adjustments that don't match the spec (tagged with *)

TxOut

• makeTxOut (Proxy era) -> mkBasicTxOut

• getField @"value" txOut -> txOut ^. valueTxOutL

• _ -> txOut ^. compactValueTxOutL

• getCoin -> txOut ^. coinTxOutL

• getTxOutAddr txOut -> txOut ^. addrTxOutL

• getTxOutEitherAddr txOut -> txOut ^. addrEitherTxOutL

• getTxOutCompactAddr txOut -> txOut ^. compactAddrTxOutL

• getTxOutBootstrapAddress txOut -> txOut ^. bootAddrTxOutF

• getField @"datahash" txOut -> txOut ^. dataHashTxOutL (*)

• getField @"referenceScript" txOut -> txOut ^. referenceScriptTxOutL

• _ -> txOut ^. dataTxOutL

• _ -> txOut ^. datumTxOutL

TxBody

• getField @"inputs" txBody -> txBody ^. inputsTxBodyL

• getField @"outputs" txBody -> txBody ^. outputsTxBodyL

• getField @"txfee" txBody -> txBody ^. feeTxBodyL (*)

• getField @"auxiliaryData" txBody -> txBody ^. auxDataHashTxBodyL (*)

• getAllInputs txBody -> txBody ^. allInputsTxBodyF

• getField @"minted" txBody -> txBody ^. mintedTxBodyF

• getField @"wdrls" txBody -> txBody ^. wdrlsTxBodyL

• getField @"ttl" txBody -> txBody ^. ttlTxBodyL

• getField @"update" txBody -> txBody ^. updateTxBodyL

• getField @"certs" txBody -> txBody ^. certsTxBodyL

• getField @"vldt" txBody -> txBody ^. vldtTxBodyL

• getField @"mint" txBody -> txBody ^. mintTxBodyL

• getField @"collateral" txBody -> txBody ^. collateralInputsTxBodyL (*)

• getField @"reqSignerHashes" txBody -> txBody ^. reqSignerHashesTxBodyL

• getField @"scriptIntegrityHash" txBody -> txBody ^. scriptIntegrityHashTxBodyL

• getField @"txnetworkid" txBody -> txBody ^. networkIdTxBodyL (*)

• getField @"sizedOutputs" txBody -> txBody ^. sizedOutputsTxBodyL

• getField @"referenceInputs" txBody -> txBody ^. referenceInputsTxBodyL

• getField @"totalCollateral" txBody -> txBody ^. totalCollateralTxBodyL

• getField @"collateralReturn" txBody -> txBody ^. collateralReturnTxBodyL

• getField @"sizedCollateralReturn" txBody -> txBody ^. sizedCollateralReturnTxBodyL

Tx

• getField @"body" tx -> tx ^. bodyTxL
• getField @"wits" tx -> tx ^. witsTxL
• getField @"auxData" tx -> tx ^. auxDataTxL
• getField @"size" tx -> tx ^. sizeTxF
• getField @"isValid" tx -> tx ^. isValidTxL

Nested instances are no longer needed, since composition of lenses works much better:

• getField @"addrWits" tx -> tx ^. witsTxL . addrWitsL
• getField @"scriptWits" tx -> tx ^. witsTxL . scriptsWitsL
• getField @"bootWits" tx -> tx ^. witsTxL . bootAddrWitsL
• getField @"txdatahash" tx -> tx ^. witsTxL . datsWitsL . to unTxDats

Witnesses

• getField @"addr" txWits -> txWits ^. addrWitsL
• getField @"bootAddr" txWits -> txWits ^. bootAddrWitsL
• getField @"script" txWits -> txWits ^. scriptWitsL
• getField @"dats" txWits -> txWits ^. datsWitsL
• getField @"rdmrs" txWits -> txWits ^. rdmrsWitsL

Removal of constraint type synonyms

All of the Trans*, Uses* constraints have been removed as they are fully
replaced by the above hierarchy

Names removed and their closest replacements:

• BlockAnn -> EraTx
• ChainData - no replacemnt, this was just an adhoc collection of classes
  SerialisableData,
• AnnotatedData -> FromCBOR (Annotator t) + ToCBOR t
• SerialisableData -> FromCBOR t + ToCBOR t
• WellFormed -> EraTx

Val class

Addition of two class methods that allow efficient manipulation of compact coin
representation: injectCompact and modifyCompactCoin, without uncompacting
the actual type t for which the Val instance is being defined

Data type renaming

All concrete types for every era where either inconsistent or clashing with the
core type families:

• Before this PR:

Core	Shelley	ShelleyMA	Alonzo	Babbage
Tx	Tx	--	ValidatedTx	--
TxBody	TxBody	MATxBody	TxBody	TxBody
TxOut	TxOut	--	TxOut	TxOut
AuxiliaryData	AuxiliaryData	MAAuxiliaryData	AuxiliaryData	--
Witnesses	WitnessSetHKD	--	TxWitness	--
Script	MultiSig	Timelock	Script	--
Value	Coin	Value	--	--
PParams	PParams	--	PParams	PParams
PParamsDelta	PParamsUpdate	--	PParamsUpdate	PParamsUpdate

• After this PR

Core	Shelley	ShelleyMA	Alonzo	Babbage
Tx	ShelleyTx	--	AlonzoTx	--
TxBody	ShelleyTxBody	MATxBody	AlonzoTxBody	BabbageTxBody
TxOut	ShelleyTxOut	--	AlonzoTxOut	BabbageTxOut
AuxiliaryData	ShelleyAuxiliaryData	MAAuxiliaryData	AlonzoAuxiliaryData	--
Witnesses	WitnessSetHKD	--	TxWitness	--
Script	MultiSig	Timelock	AlonzoScript	--
Value	Coin	MaryValue	--	--
PParams	ShelleyPParams	--	AlonzoPParams	BabbagePParams
PParamsUpdate	ShelleyPParamsUpdate	--	AlonzoPParamsUpdate	BabbagePParamsUpdate

Module restructure:

Concrete era data type (eg. BabageEra) is no longer defined at the top level
module (eg. Cardano.Ledger.Babbage), but rather in a new private .Era module
(eg. Cardano.Ledger.Babbage.Era). It is later re-exported from the top level
module together with all of the orphan instances. This allows us to define all
of the relevant instances in their respective modules (eg. EraTxBody is defined
in the TxBody module)

Other minor changes

• Defined NFData for Sized
• Move module Cardano.Ledger.Shelley.Address.Bootstrap
  (cardano-ledger-shelley) -> Cardano.Ledger.Keys.Bootstrap (cardano-ledger-core)
• Moved langsUsed into tests, since that was the only use site.

[JaredCorduan]

Colossal cleanup!! Thank you so much for putting in all this effort to make our codebase more friendly!

I'm happy with all the changes. I made a handful of comments, but nothing of real importance.

🚀

[JaredCorduan]

@lehins thank you also for the amazing PR notes. Not only is it very helpful for reviewers, it leaves a great trail for others later down the line. (and for myself too, later, when I might have forgotten some details). 💪