create-testnet-data: use experimental API and make arguments era specific

[smelc]

Changelog

- description: |
    More fine grained controls of eras for create-testnet-data. Forbid creating DReps and CC in babbage.

    Remove the `create-testnet-data` altogether in eras earlier than Babbage
# uncomment types applicable to the change:
  type:
  # - feature        # introduces a new feature
  - breaking       # the API has changed in a breaking way
  # - compatible     # the API has changed but is non-breaking
  # - optimisation   # measurable performance improvements
  # - refactoring    # QoL changes
  # - bugfix         # fixes a defect
  # - test           # fixes/modifies tests
  # - maintenance    # not directly related to the code
  # - release        # related to a new release preparation
  # - documentation  # change in code docs, haddocks...

Context

Fixes #962

How to trust this PR

Look at the changes in golden files

Checklist

• Commit sequence broadly makes sense and commits have useful messages
• Self-reviewed the diff

[palas]

Looks good. I left a couple of comments on things that caught my eye, just minor things. Feel free to address them if you think they make sense, but I leave it to your judgment

[palas]

It doesn't derive anymore? It is odd because Era does have a Show instance (here), maybe after moving the parameter it will?

[smelc]

@palas> it doesn't derive anymore because there's an existential type in GenesisCreateTestnetDataCmdArgs now:

[palas]

Making it unrepresentable on the type would be safer. But this is probably much simpler, so it seems a reasonable trade-off

[carbolymer]

Those inlined cases are not very readable, can you move them to a values in where, and just reference them from here?

[carbolymer]

I'd also use Ord (Some Era) instead of pattern matching on each era. I mean:

if someEra < Some Exp.ConwayEra 
  then pure 0 
  else pNumCommittee

and

if someEra < Some Exp.ConwayEra
  then pure $ DRepCredentials OnDisk 0
  else pNumDReps

Rationale: less maintenance burden when introducing new eras - you won't have to change each usage site of case.

[smelc]

I'd also use Ord (Some Era) instead of pattern matching on each era. I mean:

@carbolymer> I'm not sure this is really better, because there are at most two experimental eras at any point in time and when we delete one era, the index of the other era (the one being kept) shrinks from 1 to 0. So I wouldn't rely on the ordering here. I think pattern matching makes the intent more explicit.

[palas]

Ah, one more thing, probably say what got broken in the change log. (That some eras no longer have the command.)

[carbolymer]

Why not Some Exp.Era instead?