Cardano Updates

The crate gated `serde` behind the `json` feature, but used
`serde::{Serialize, Deserialize}` unconditionally across the era
models, so building with `default-features = false` failed to
compile. Since `pallas-codec` and `pallas-crypto` already depend on
serde unconditionally, gating it here gave no real benefit anyway.

The `json` feature now gates only `serde_json`. A CI job checks
pallas-primitives in isolation, since a workspace-wide
`--no-default-features` build unifies the feature back on.

Fixes #740

Co-authored-by: Claude Opus 4.7 (1M context) <[email protected]>

The crate gated `serde` behind the `json` feature, but used
`serde::{Serialize, Deserialize}` unconditionally across the era
models, so building with `default-features = false` failed to
compile. Since `pallas-codec` and `pallas-crypto` already depend on
serde unconditionally, gating it here gave no real benefit anyway.

The `json` feature now gates only `serde_json`. A CI job checks
pallas-primitives in isolation, since a workspace-wide
`--no-default-features` build unifies the feature back on.

Fixes #740

Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>

* fix: correct inverted native script timelock comparisons

`eval_native_script` evaluated both timelock opcodes with inverted
comparisons. `InvalidBefore(n)` was satisfied when the validity
interval started before slot `n` (should be at or after), and
`InvalidHereafter(n)` when it ended after `n` (should be at or
before), per mary-ledger.pdf p.20.

Adds unit tests covering both opcodes and nested combinators.

Fixes #718

Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>

* style: apply rustfmt

Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>

---------

Co-authored-by: Claude Opus 4.7 (1M context) <[email protected]>

Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>

`eval_native_script` evaluated both timelock opcodes with inverted
comparisons. `InvalidBefore(n)` was satisfied when the validity
interval started before slot `n` (should be at or after), and
`InvalidHereafter(n)` when it ended after `n` (should be at or
before), per mary-ledger.pdf p.20.

Adds unit tests covering both opcodes and nested combinators.

Fixes #718

Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>

This allows to enforce the computation of the DMQ message id based on the DMQ payload.