test(plutus): speed up PlutusV3 builtins tests

[mkoura]

test_plutusv3_builtins_new_cost was the longest test in the Leios testrun (220 min). It was also picked up only 86 min into the run, so it finished an hour after all other tests, with 19 of 20 workers idle. The builtins tests ran one script at a time, and each script waited for its own funding tx and its own mint tx.

Schedule new_cost first

Mark test_plutusv3_builtins_new_cost with order(5), so it starts in the first wave of long tests instead of near the end.

Fund the token issuer in batches

• Fund the issuer for up to 30 scripts in a single tx, instead of one funding tx per script. The tests with the old cost model, where mints fail already at build time, spent nearly all their time on funding.
• Fund the payment address once per test, for all chunks.
• Cache the addresses per variant, so tests running in parallel on the same cluster instance don't share them.
• Check the outputs of the funding tx instead of the issuer address balance.

Confirm mints per chunk

• Submit the mint txs without waiting for them to appear on chain, and confirm all mints of a chunk together. That is one wait per chunk instead of one per script.
• Expected failures are still caught at build or submit time, in the script's own subtest.
• Each submitted mint is checked in its own _minted subtest.
• Txs that didn't make it to the chain are resubmitted, but only after a round where none got confirmed, as not all of them fit into one block. Only such rounds count against the attempts limit.
• Tx file names include the script index, so scripts with the same file stem don't overwrite each other's pending txs.

Results (Leios testrun, 20 workers, 9 cluster instances)

	before	after
Testrun wall time	317 min	245 min
new_cost picked up / finished	86 / 317 min	0 / 54 min
new_cost	220 min	48 min
old[batch5]	33 min	2.2 min
old[batch6]	101 min	4.4 min

The run is now evenly packed: all workers are busy until the end, and there is no single-test tail.

[Copilot]

Copilot review overview

🟢 Approval recommended

The batching and confirmation logic is consistent with existing test infrastructure and no blocking issues were found.

Review effort: Balanced
Findings: None

What changed in this PR

Optimizes PlutusV3 builtin tests to reduce Leios test-run duration.

Changes:

• Batches issuer funding and mint confirmation.
• Uses variant-specific cached addresses and unique transaction names.
• Schedules the longest test earlier.

File	Description
cardano_node_tests/​tests/​tests_plutus_v3/​test_mint_build.py	Adds batched funding, asynchronous mint confirmation, and early scheduling.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.