fix(integration): fix partitions,permissions integration tests

[numinnex]

Fixes integration tests for partitions (excluding those that use consumer_group) & permissions auth tests.

[codecov]

Codecov Report

❌ Patch coverage is 19.58763% with 858 lines in your changes missing coverage. Please review.
✅ Project coverage is 74.21%. Comparing base (b52e40e) to head (cb679ae).

Files with missing lines	Patch %	Lines
core/server-ng/src/dispatch.rs	0.00%	279 Missing ⚠️
core/partitions/src/iggy_partition.rs	7.45%	268 Missing and 5 partials ⚠️
core/server-ng/src/responses.rs	0.00%	82 Missing ⚠️
core/shard/src/lib.rs	36.36%	59 Missing and 4 partials ⚠️
core/shard/src/router.rs	0.00%	22 Missing ⚠️
core/server-ng/src/main.rs	0.00%	21 Missing ⚠️
core/server-ng/src/bootstrap.rs	0.00%	20 Missing ⚠️
core/server-ng/src/users.rs	0.00%	17 Missing ⚠️
core/server-ng/src/wire.rs	0.00%	17 Missing ⚠️
core/server_common/src/send_messages2.rs	38.88%	10 Missing and 1 partial ⚠️
... and 13 more

❌ Your patch check has failed because the patch coverage (19.58%) is below the target coverage (50.00%). You can increase the patch coverage or adjust the target coverage.

Additional details and impacted files

@@             Coverage Diff              @@
##             master    #3487      +/-   ##
============================================
- Coverage     74.70%   74.21%   -0.49%     
  Complexity      937      937              
============================================
  Files          1258     1259       +1     
  Lines        125131   125969     +838     
  Branches     100805   101688     +883     
============================================
+ Hits          93475    93490      +15     
- Misses        28658    29418     +760     
- Partials       2998     3061      +63     

Components	Coverage Δ
Rust Core	75.13% <19.58%> (-0.57%)	⬇️
Java SDK	58.57% <ø> (ø)
C# SDK	71.40% <ø> (-0.71%)	⬇️
Python SDK	88.88% <ø> (ø)
PHP SDK	84.29% <ø> (ø)
Node SDK	91.22% <ø> (-0.13%)	⬇️
Go SDK	40.36% <ø> (ø)

Files with missing lines	Coverage Δ
core/binary_protocol/src/consensus/header.rs	79.72% <100.00%> (+0.03%)	⬆️
core/consensus/src/observability.rs	41.47% <100.00%> (ø)
core/consensus/src/plane_helpers.rs	91.46% <100.00%> (+1.14%)	⬆️
core/message_bus/src/socket_opts.rs	82.35% <ø> (-0.51%)	⬇️
core/metadata/src/impls/metadata.rs	38.90% <100.00%> (+0.20%)	⬆️
core/sdk/src/websocket/websocket_client.rs	71.73% <ø> (ø)
core/server-ng/src/partition_helpers.rs	61.66% <100.00%> (+0.62%)	⬆️
core/server-ng/src/pat.rs	0.00% <ø> (ø)
core/journal/src/prepare_journal.rs	86.64% <95.00%> (+0.39%)	⬆️
core/message_bus/src/config.rs	81.57% <66.66%> (-1.09%)	⬇️
... and 21 more

... and 43 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[numinnex]

/ready

[numinnex]

/ready

[hubcio]

a few findings land on code this PR didn't change, so they're here rather than as line comments:

• core/sdk/src/websocket/websocket_client.rs (around 700-716): the WS vsr send_raw reads the response header and body with no timeout while holding the stream lock, which is the exact lockstep wedge the TCP client just got fixed (RESPONSE_READ_TIMEOUT plus drop-on-timeout). if the server loses a reply the WS client hangs forever. the new sdk/messages.rs test runs over WebSocket too, so this path is now exercised. it's actually worse than the old TCP case - the WS read runs in the caller's task, not a spawned one, so there's no post-task stream drop. mirror the TCP timeout plus stream-drop on WS (and check QUIC).

• core/server-ng/src/responses.rs:294 and :597: the POLL_MESSAGES arm in build_non_replicated_response and the NonReplicatedResponse::EmptyPolledMessages variant look dead now - handle_poll_messages has its own explicit dispatch arm, so the fallback never reaches POLL. ~15 lines plus a never-served empty-poll path can go.

• core/server_common/src/send_messages2.rs:465: owned_message is #[allow(dead_code)] with zero callers, and it happens to be the one place with the correct flat-base_timestamp reconstruction. either wire it into build_polled_messages_body (which would also fix the timestamp bug) or delete it.

[numinnex]

/ready

[numinnex]

/ready