feat(bench): add vault transactions to public mix without zones

[decofe]

Adds vault deposits and withdrawals to public-mix, with deployment, funding, and per-user share seeding, so vault traffic can be benchmarked without zone transactions or zone setup. Based on #7569: vaults make up 20% of submitted transactions (12% deposits, 8% withdrawals), while the remaining 80% preserves the transfer/memo/mint/MPP ratio of 25/40/1/15. Vault workload templates use expiring nonces, and category shares remain constant across account counts.

Exposes general-gas-limit in the multi-region manual workflow so non-payment traffic can use a larger gas budget. Manual runs retain four TIP20 tokens to stay within GitHub's 25-input limit. The runner is pinned to the tested command replay fix, which is required to avoid repeating fixture setup after result-upload failures.

Validation: preset rendering was checked at 1, 3, and 1,000 users; workflow YAML, dispatch input count, input wiring, and git diff --check pass. The multi-region comparison and infrastructure teardown succeeded: 4,809 included TPS baseline / 4,642 feature, with 12,467 / 13,562 reported transaction failures. The saved genesis and runtime configuration both confirm a 5B non-payment gas cap. The run used a 5B total gas cap, 20% vault traffic, 10 validators across four GCP regions, 100 GiB state, a 50k TPS target, and one 90-second baseline/feature pair. The 50k TPS target was not achieved; one pair does not establish a performance regression.

Prompted by: @shekhirin