Split Global_slot.t into since-genesis and since-hard-fork types

[psteckler]

Split Mina_numbers.Global_slot.t into Global_slot_since_genesis.t and Global_slot_since_hard_fork.t. Add new type Global_slot_span.t for vesting periods, also used as for an argument to add and sub. Although all these types as an unsigned 32-bit integer, none is compatible with the others, so they can't be confused in the code.

The existing uses of Global_slot.t have been changed to the ...since_genesis.t type. Some of them very likely need to become uses of ...since_hard_fork.t.

For V1 Signed_commands, introduce Global_slot_legacy, which is identical to the old Global_slot.

There were places in the code where slots were being used both as slots, as such, and as spans. In this PR, that's handled by converting the values via uint32. Probably that code should be cleaned up to make those roles distinct.

Closes #13058.

[psteckler]

!ci-build-me

[psteckler]

!ci-build-me

[jdsteinhauser]

That was a huge PR! Looks good,

[rbonichon]

@psteckler Would you be able to open issues wrt to spans/slots uses and wrt to this remark

Some of them very likely need to become uses of ...since_hard_fork.t.

[psteckler]

@psteckler Would you be able to open issues

#13154 #13155

[psteckler]

!ci-build-me

[rbonichon]

Few minor suggestions. LGTM

[rbonichon]

I'm not sure I get why we need the conversion for t' and not for t as they have the same type so I was expecting this treatment to be symmetric (old code had no conversion whatsoever)

[psteckler]

T.sub takes a slot and a span, so t'.slot_number needed to be converted to a span.

T.sub used to take a slot and another slot, but I think it makes more sense to subtract a span, in the same way Time does.

But ( - ) isn't used anywhere, so I'm going to remove it. Its uses were replaced by diff_slots, which takes the slots from two ts, and returns a span.

[rbonichon]

I don't know how many times this pattern occur but is it worth having succ in Global_slot_span too?

[psteckler]

It's there already. But this is an instance where a span is converted to a slot via uint32. The succ could be done in the span, the uint32, or the slot, and I chose the uint32. I'll change it to do it in the span.

[psteckler]

!ci-build-me

[ghost-not-in-the-shell]

!ci-build-me

[mitschabaude]

!ci-build-me

[mitschabaude]

!ci-build-me

[mitschabaude]

!ci-build-me

[deepthiskumar]

!ci-build-me

[psteckler]

!ci-build-me

[psteckler]

!ci-build-me

[psteckler]

!ci-build-me

[psteckler]

!ci-build-me

[deepthiskumar]

!approved-for-mainnet

[deepthiskumar]

The following version change from mainnet type is actually to retain the old global slot type. Rest of the version changes are for V2 types or V1 types introduced for zkApps

In file: src/lib/mina_base/signed_command_payload.ml
--
  | Versioned types changed at path: Common.Stable.V1.t
  | Was: type t = (Currency.Fee.Stable.V1.t, Public_key.Compressed.Stable.V1.t, Token_id.Stable.V1.t, Account_nonce.Stable.V1.t, Global_slot.Stable.V1.t, Memo.Stable.V1.t) Poly.Stable.V1.t[@@deriving bin_io]
  | Now: type t = (Currency.Fee.Stable.V1.t, Public_key.Compressed.Stable.V1.t, Token_id.Stable.V1.t, Account_nonce.Stable.V1.t, Global_slot_legacy.Stable.V1.t, Memo.Stable.V1.t) Poly.Stable.V1.t [@@deriving bin_io]