CI on this PR caught a regression: CW1155toERC1155PointerTest's "should
not transfer an NFT if not owned" reads res.code from executeWasm and
expects it to be non-zero (the wasm execution should fail when the
sender doesn't own the tokens). Under -b block, res.code is the
DeliverTx code (post-execution); under -b sync it's the CheckTx code
(mempool acceptance). For a tx that passes basic checks but fails in
DeliverTx, -b sync returns 0 and the test wrongly thinks the tx
succeeded.

executeWasm and associateWasm have callers that read post-execution
fields (code, events, logs) from the response, so they can't migrate
to the -b sync pattern without also adding a "submit then poll for tx
result" step. Revert these two; leave the truly response-agnostic
helpers (bankSend, fundSeiAddress, associateKey, passProposal vote)
on -b sync. The complex helpers are deferred to the same follow-up
PR that addresses storeWasm, proposeParamChange, and the other
event-parsing callers.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

CI on this PR caught a sequence mismatch in SeiSoloTest's
"Claim CW20 Tester / before all": fundSeiAddress submits with -b sync
and waitForBlocks(1) returns when the next block lands. That next
block can be empty (the submitted tx is still in mempool and lands
one block later), so the chain hasn't advanced the funder's account
sequence yet. The follow-up storeWasm queries the funder's account,
gets the stale sequence, signs with it, and is rejected when the
funder tx finally lands a block later and bumps the sequence.

Default to 2 blocks instead of 1: closes the empty-next-block race
without reintroducing a fixed sleep. Callers no longer pass an
explicit block count — the helper takes care of the safety margin.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Two more lib.js helpers were still using --broadcast-mode=block. Under
Autobahn the executeBlock path doesn't fire EventDataTx, so -b block
hangs to its 60s timeout on every call.

evmSend (used by fundAddress, called from setupSigners and many test
file before-hooks): convert to -b sync + waitForBlocks. Callers read
only the txhash from the output, which is present in either format,
so no other adjustment needed. This unblocks EVMPrecompileTest and
EVMGigaTest's `transfer to non-existent contract with data` (latter
verified passing in 126ms after the fix).

proposeParamChange: convert to -b sync. The tx response under -b sync
no longer carries deliver_tx events, so we can't read proposal_id from
the submit_proposal event. Poll gov state instead (max-id-before vs
max-id-after) — autobahn doesn't run a Cosmos-side tx indexer so
seid q tx <hash> isn't an option. New maxProposalId helper handles the
"no proposals exist yet" case explicitly via try/catch; using a shell
`|| echo '{...}'` fallback breaks because the docker-exec wrapper
double-quotes the inner command. Verified by EVMCompatabilityTest
test #3 (`should reproduce mismatch by changing param`) passing in
47s after the fix (was timing out at 120s with -b block).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Last lib.js helper still using --broadcast-mode=block. Same Autobahn
incompatibility as evmSend / proposeParamChange: executeBlock doesn't
fire EventDataTx, so -b block hangs to its 60s timeout on every call,
which blocked ERC20toNativePointerTest's before-all hook.

The original code read the new denom from the create_denom event in
the tx response. Under -b sync we don't get deliver_tx events, but
the tokenfactory denom is deterministic — `factory/<creator>/<subdenom>`
where creator is the bech32 of the --from key. Construct it locally
from getKeySeiAddress instead.

Verified: ERC20toNativePointerTest passes 14/14 in 17s after the fix
(previously 0/1, hung 60s+ on the before-all hook).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Gov Precompile section's before-hook had its own inline
'seid tx gov submit-proposal param-change ... -b block' which hangs
60s under Autobahn. Replace with the lib.js proposeParamChange helper
(now -b sync + poll gov state).

EVMPrecompileTest goes from 9/11 to 10/11 — only the Wasm Precompile
section still fails (out of scope; wasm being decommed).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Strip comments that described the old -b block behavior or contrasted
-b sync against it. Reader of the current code doesn't need to know
what was replaced; comments now describe pure invariants
(empty-next-block race, deterministic denom format, diff-against-snapshot
proposal identification, seid's empty-gov-set exit code).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Read title/description/changes/is_expedited from the JSON file directly
rather than duplicating them inline. deposit/fees stay inline (they're
not part of the proposal spec; they're CLI-level args).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

…aramChange

After polling identifies the new proposal id, fetch the proposal and
assert its title matches what we submitted. Closes the sharp edge where
a concurrent submission (or DeliverTx silent failure followed by an
unrelated submission) would otherwise cause us to return someone else's
id.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Keep original command/mint_command/send_command variable names and
inline-execute pattern. Only changes vs main are: derive token_denom
locally instead of from the create_denom event, swap -b block for
-b sync, and waitForBlocks between submissions.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>