Fix SLO KV consumer and process failure gate

[KirillKurdyukov]

Summary

Consume the basic KV checks from
ydb-platform/ydb-java-examples#69 at the exact public
revision 68c06dfa5f09dea42613dc5eaffb0e4c456cfd51.

Keep the four existing Query, JDBC, Spring Data JDBC and Spring Data JPA
workloads and build each role against its own SDK checkout. Remove the
baseline invocation's current-image fallback so a failed baseline source
build cannot silently run the current SDK instead. Enable the existing
init@v2 process-error gate with fail_on_workload_error: true.

This change touches only .github/workflows/slo.yml. It does not change SDK
code, retry/deadline settings, the build helper API or metrics.

Verification

• Parsed the final workflow and checked the exact examples SHA, all four
  matrix rows, separate SDK roots, absent baseline fallback and true process
  gate. The remainder is semantically identical to the base workflow.
• Built all eight current/baseline source images with the original helper.
  Both role SDK library trees use d854009; this proves consumption, not a
  different-version benchmark.
• Ran eight healthy native processes with isolated lifecycles and sixteen
  negative processes covering missing/stale/wrong payload and final SDK
  errors across the four adapters. Healthy processes exit 0; negatives exit
  1 and retain their cause.
• Inspected the exact published Action v2 process gate. Ran whitespace and
  helper shell-syntax checks. The one-file source-publication review passed.
• Examples CI passed on JDK 8, 11 and 17 on unchanged-source attempt 2:
  https://github.com/ydb-platform/ydb-java-examples/actions/runs/38666103274
  Attempt 1 hit a port collision in unchanged query-example setup; the
  cancelled JDK jobs are not counted as passes.

Hosted SDK SLO has not yet run this consumer. Deliberately induced native
retry recovery and a two-version performance comparison are not claimed.
The examples prerequisite and this consumer remain separate, unmerged PRs.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 75.48%. Comparing base (d854009) to head (a11a771).

Additional details and impacted files

@@             Coverage Diff              @@
##             master     #772      +/-   ##
============================================
+ Coverage     75.47%   75.48%   +0.01%     
- Complexity     3755     3756       +1     
============================================
  Files           396      396              
  Lines         16842    16842              
  Branches       1780     1780              
============================================
+ Hits          12711    12714       +3     
+ Misses         3515     3513       -2     
+ Partials        616      615       -1     

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.