van Rossem: backport new Plutus builtins' costing to V1/V2 cost models

[scarmuega]

Context

Per van Rossem (Cardano protocol v11), the new Plutus builtins introduced via CIPs 109, 132, 133, 138, and 153 — including ripemd_160, expModInteger, dropList, lengthOfArray, listToArray, indexArray, plus the bytestring/bit ops 75–85 — are backported to all three Plutus language versions (V1, V2, V3), not only V3.

Sources:

• https://forum.cardano.org/t/mar-5-2026-parameter-committee-triweekly-meeting-notes/154363
• https://forum.cardano.org/t/plutus-cost-model-increases-for-post-vanrossem-potentially-breaking-contracts/153744

Current state

Pallas now uses amaru-uplc 0.1.0 as the phase-2 backend (since a7b5a86, "switch from txpipe uplc-turbo fork to pragma-org amaru-uplc"). In that crate:

• src/builtin/default_function.rs declares all new variants (Ripemd_160 = 86, ExpModInteger = 87, DropList = 88, LengthOfArray = 89, ListToArray = 90, IndexArray = 91).
• src/machine/cost_model/builtin_costs/builtin_costs_v3.rs has costing entries for all of them (33 references).
• src/machine/cost_model/builtin_costs/builtin_costs_v1.rs and builtin_costs_v2.rs have zero references to the new builtins.

Verified via:

$ grep -c "ripemd_160\|exp_mod_integer\|drop_list\|length_of_array\|list_to_array\|index_array" \
    src/machine/cost_model/builtin_costs/builtin_costs_v{1,2,3}.rs
v1: 0
v2: 0
v3: 33

Upstream tracking

Filed upstream as pragma-org/uplc#48. The bulk of the change lives there; this issue tracks the pallas-side bump once upstream lands.

Required change

Either:

1. Coordinate upstream with pragma-org/uplc (see #48) to extend builtin_costs_v1.rs and builtin_costs_v2.rs with the same builtin set, then bump the dependency in pallas-validate/Cargo.toml.
2. Or, fork the cost-model files locally in pallas if the upstream timeline doesn't match.

Without this, V1/V2 cost-model arrays decoded from post-v11 PParamsUpdate proposals will fail to map to costing entries, breaking phase-2 evaluation for V1/V2 scripts.

Why this is blocking

Tracked here as it blocks pallas's van Rossem readiness even though the bulk of the change lives in the amaru-uplc crate.