fix(amber): correct WorkerSpec input port to stop flaky timeouts

[Ma77Ball]

What changes were proposed in this PR?

• WorkerSpec's input-port AssignPortRequest now passes an empty storageUris (List()) matching its empty partitionings, satisfying InputManager.addPort's size invariant, so the input port registers and the three worker tests pass deterministically.
• AsyncRPCServer.invokeMethod now re-throws Errors (e.g. failed assertions) after replying to the sender, instead of swallowing them; normal Exceptions are still returned to the sender unchanged, so genuine invariant violations surface loudly rather than degrading into flaky timeouts.

Any related issues, documentation, discussions?

Closes: #6036

How was this PR tested?

• Run sbt "WorkflowExecutionService/testOnly *WorkerSpec"; expect Tests: succeeded 3, failed 0.
• Diagnostic: grep the run log for assertion failed. Before this fix the count is 3 (the assertion fired and was swallowed every run); after it is 0. Verified locally under Java 17.
• The full amber suite is heavy and flaky to run locally, so the engine-wide AsyncRPCServer change is left to the amber CI job to exercise across all RPC handlers.

Was this PR authored or co-authored using generative AI tooling?

Co-authored with Claude Opus 4.8 in compliance with ASF

[github-actions]

Automated Reviewer Suggestions

Based on the git blame history of the changed files, we recommend the following reviewers:

• Contributors with relevant context: @Yicong-Huang
  You can notify them by mentioning @Yicong-Huang in a comment.

[github-actions]

✅ No material benchmark regressions detected

🟢 5 better · 🔴 0 worse · ⚪ 10 noise (<±5%) · 0 without baseline

Compared against main 1073b22 benchmarked on this same runner, so the delta is largely free of cross-runner hardware noise. The "7d avg" column still reflects the gh-pages dashboard. Treat <±5% as noise unless repeated.

Dashboard · Run

	config	throughput	MB/s	latency	max Δ latest / 7d
🟢	bs=10 sw=10 sl=64	591	0.361	16,164/22,947/22,947 us	🟢 -10.7% / 🔴 +55.2%
🟢	bs=100 sw=10 sl=64	1,242	0.758	78,658/98,798/98,798 us	🟢 -7.0% / 🟢 +25.3%
⚪	bs=1000 sw=10 sl=64	1,405	0.858	706,210/776,519/776,519 us	⚪ within ±5% / 🟢 +37.4%

Baseline details

Latest main 1073b22 from same runner

config	metric	PR	latest main	7d avg	Δ latest	Δ 7d
bs=10 sw=10 sl=64	throughput	591 tuples/sec	583 tuples/sec	786.27 tuples/sec	+1.4%	-24.8%
bs=10 sw=10 sl=64	MB/s	0.361 MB/s	0.356 MB/s	0.48 MB/s	+1.4%	-24.8%
bs=10 sw=10 sl=64	p50	16,164 us	17,232 us	12,495 us	-6.2%	+29.4%
bs=10 sw=10 sl=64	p95	22,947 us	25,697 us	14,784 us	-10.7%	+55.2%
bs=10 sw=10 sl=64	p99	22,947 us	25,697 us	18,468 us	-10.7%	+24.3%
bs=100 sw=10 sl=64	throughput	1,242 tuples/sec	1,226 tuples/sec	991.49 tuples/sec	+1.3%	+25.3%
bs=100 sw=10 sl=64	MB/s	0.758 MB/s	0.748 MB/s	0.605 MB/s	+1.3%	+25.3%
bs=100 sw=10 sl=64	p50	78,658 us	80,848 us	100,929 us	-2.7%	-22.1%
bs=100 sw=10 sl=64	p95	98,798 us	106,221 us	106,894 us	-7.0%	-7.6%
bs=100 sw=10 sl=64	p99	98,798 us	106,221 us	114,085 us	-7.0%	-13.4%
bs=1000 sw=10 sl=64	throughput	1,405 tuples/sec	1,440 tuples/sec	1,023 tuples/sec	-2.4%	+37.4%
bs=1000 sw=10 sl=64	MB/s	0.858 MB/s	0.879 MB/s	0.624 MB/s	-2.4%	+37.4%
bs=1000 sw=10 sl=64	p50	706,210 us	698,853 us	983,835 us	+1.1%	-28.2%
bs=1000 sw=10 sl=64	p95	776,519 us	741,459 us	1,023,777 us	+4.7%	-24.2%
bs=1000 sw=10 sl=64	p99	776,519 us	741,459 us	1,053,883 us	+4.7%	-26.3%

Raw CSV

config_idx,batch_size,schema_width,string_len,num_batches,total_ms,total_tuples,total_bytes,tuples_per_sec,mb_per_sec,lat_p50_us,lat_p95_us,lat_p99_us
0,10,10,64,20,338.22,200,128000,591,0.361,16163.87,22946.85,22946.85
1,100,10,64,20,1610.60,2000,1280000,1242,0.758,78658.20,98797.54,98797.54
2,1000,10,64,20,14231.84,20000,12800000,1405,0.858,706209.65,776519.38,776519.38

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 56.87%. Comparing base (c3161f7) to head (6d39dfc).
⚠️ Report is 70 commits behind head on main.

Additional details and impacted files

@@             Coverage Diff              @@
##               main    #6037      +/-   ##
============================================
+ Coverage     54.50%   56.87%   +2.37%     
+ Complexity     2915     1901    -1014     
============================================
  Files          1108      843     -265     
  Lines         42807    34646    -8161     
  Branches       4604     3480    -1124     
============================================
- Hits          23332    19705    -3627     
+ Misses        18119    13986    -4133     
+ Partials       1356      955     -401     

Flag	Coverage Δ		*Carryforward flag
access-control-service	70.44% <ø> (ø)		Carriedforward from 12c1107
agent-service	34.36% <ø> (ø)		Carriedforward from 12c1107
amber	68.17% <ø> (+11.65%)	⬆️
computing-unit-managing-service	1.65% <ø> (ø)		Carriedforward from 12c1107
config-service	57.35% <ø> (ø)		Carriedforward from 12c1107
file-service	58.59% <ø> (ø)		Carriedforward from 12c1107
frontend	48.27% <ø> (ø)		Carriedforward from 12c1107
pyamber	90.20% <ø> (ø)		Carriedforward from 12c1107
python	90.76% <ø> (ø)		Carriedforward from 12c1107
workflow-compiling-service	58.69% <ø> (ø)		Carriedforward from 12c1107

*This pull request uses carry forward flags. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[Yicong-Huang]

how come non errors can reach here? that's a bigger problem...

with your change, this "non-error" is omitted silently...

can we at least log warning when it happens?

[Ma77Ball]

Dropped the throw since it can take down the actor loop. The Error was an AssertionError from the test's mismatched addPort args; it now just logs a warning so it's not silent.

[Yicong-Huang]

The src code seems to have a bigger issue. we should not make source code less robust to satisfy test cases...

[Ma77Ball]

Reverted the re-throw. Fixed the real root cause: the test passed mismatched storageUris/partitionings, now corrected in one line. AsyncRPCServer only keeps a warning log so a swallowed Error isn't silent.