Merge main into release/v3.x-new in order to prepare for v3.6.17 release

[dusan-maksimovic]

Description

Closes: #XXXX

Add a description of the changes that this PR introduces and the files that
are the most critical to review.

---

Author Checklist

All items are required. Please add a note to the item if the item is not applicable and
please add links to any relevant follow up issues.

I have...

• Targeted the correct branch
• Included the necessary unit tests
• Added/adjusted the necessary interchain tests
• Added necessary migration code for all stores that were adjusted or added
• Added a changelog entry in .changelog
• Compiled the contracts by using make compile and included content of the artifacts directory into the PR
• Regenerated front-end schema by using make schema and included generated files into the PR
• Updated the relevant documentation or specification
• Reviewed "Files changed" and left comments if necessary
• Confirmed all CI checks have passed

[claude]

Review Summary

This PR merges main into release/v3.x-new to prepare the v3.6.17 release. It bundles several already-individually-reviewed PRs (#439, #440, #442, #444, #446, #447, #448, #451). I focused the review on the substantive Rust source changes rather than the generated artifacts/schemas/ts_types.

New IBC Eureka Inflow adapter (contracts/inflow/ibc-eureka-adapter-cosmoshub)

Overall well structured and thoroughly tested (testing_standard_adapter.rs + testing_custom_adapter.rs cover instantiate validation, deposit/withdraw, admin/executor management, denom/destination allowlists, and TransferFunds including the expired-quote and insufficient-balance paths).

A few minor observations, none blocking:

• error.rs: UnauthorizedExecutor error message reads "only executors or admins can call this function", but validate_executor_caller (validation.rs) only checks the EXECUTORS map — admins are not implicitly treated as executors. Message is slightly misleading; worth aligning wording with actual behavior (or making admins bypass the executor check if that's the intent).
• execute_register_depositor deserializes capabilities straight from the caller-supplied metadata binary via cosmwasm_std::from_json. Since this call is restricted to admins already (validate_admin_caller), this isn't a security issue, just worth noting there's no test for the malformed-metadata error path.
• Nice touch documenting the deliberate unit mismatch (seconds vs. nanoseconds) between action_timeout_timestamp and fee_timeout_timestamp in both eureka.rs and at the call site in contract.rs — that kind of thing is easy to silently get wrong later.

SubmitDeployedAmount timeout (control-center)

Clean, minimal fix for the "stale DAO proposal executes an old submission" problem. The boundary test (submit_deployed_amount_timeout_test) correctly confirms block.time == timeout is accepted and block.time > timeout is rejected, matching the > comparison in contract.rs.

SwapDepositDenom (vault-cosmoshub)

Good defensive design — checks vault balance, all registered adapters' bank balances, and adapter-reported deployed positions before allowing the swap, and fails closed if a position query errors out. Test coverage is thorough (success, non-empty withdrawal queue, denom-still-held for vault/tracked/untracked adapters, shares-denom collision, provider-not-supporting-new-denom, position-query-failure, unauthorized, same-denom).

One thing worth double-checking outside the code (can't verify from the diff alone): the doc comment on swap_deposit_denom states a deposit landing "just before the swap" simply makes the swap fail rather than requiring a lock — that's only true if deposits and this execute message can't land ahead of the balance check in a way that races the check itself. Given CosmWasm's atomic per-tx execution this should hold, just flagging the assumption for a second pair of eyes given it's guarding real user funds.

lsm-hub-token-info-provider validator ratio fix

The rewrite of initialize_validator_store/initialize_validator_store_helper changes from "copy round N-1 into round N, repeated per round" to "find the last initialized round once, then stamp its validator set onto every uninitialized round up to the target." This is both a correctness fix and an efficiency win (avoids reloading/rewriting intermediate state per round). The added regression test test_initialize_validator_store_last_initialized_ten_rounds_in_the_past covers exactly the multi-round-gap scenario that looks like the original bug, and correctly asserts every intermediate round is populated.

Other changes

• marketplace: removing the unused neutron-sdk dependency and Response<NeutronMsg> typing is a reasonable cleanup since the contract doesn't appear to use Neutron-specific messages.
• hydro/liquid-collateral/ibc-eureka-adapter-cosmoshub migrate.rs: this release's migrate() is a no-op version bump; the previously-active pause/unpause migration helpers are kept around prefixed with _ for reuse in a future migration, consistent with the pattern already established elsewhere in the repo (see CLAUDE.md migration notes).
• CI workflow (populate-staging-contracts.yml) switches the staging population job from Neutron/Osmosis to Gaia/Cosmos Hub and renames the mnemonic secret — infra-only change, nothing to flag.

Checklist / housekeeping

Artifacts, schemas, and changelog entries (.changelog/v3.6.17/...) are all present and match the merged PRs. Cargo.toml workspace version was bumped to 3.6.17 and the new adapter crate was added to both Cargo.toml members and the Makefile schema/ts-codegen targets.

No security concerns identified beyond the notes above — access control (admin/executor/depositor roles) is checked consistently across the new adapter's execute handlers, and the EVM destination-address allowlist plus role separation (executors can't add new destinations themselves) limits the blast radius of a compromised executor key to funds routed to already-approved addresses.