[8.2] [MOD-18346] Backport CI runner fixes and Debian Trixie

[GuyAv46]

Combined backport of #1032 and #1031 to 8.2. Both fixes are included so CI can validate and merge them together.

• Tag coverage and benchmark EC2 runners at launch with team=ai_and_search and a normalized GitHub actor as owner.
• Run ARM unit tests on GitHub-hosted ubuntu-22.04-arm and remove ARM EC2 provisioning and inherited secrets.
• Replace active Bullseye CI jobs with gcc:14-trixie, rename the Debian installer/workflow, and update workflow dependencies. Preserve release-specific AWS settings and test options.

Fixes MOD-18346.

Validation: parsed workflow YAML, checked local reusable-workflow paths and job dependencies, verified no active Bullseye jobs remain, compiled owner-tag Python, checked EC2 launch tags, and ran shell syntax/whitespace checks. Runtime CI is pending.

---

Note

Low Risk
Changes are limited to GitHub Actions workflows and unit test fixtures; no runtime vector-index or auth logic is modified.

Overview
This backport updates CI infrastructure and stabilizes FP16 unit tests on hosted ARM.

ARM CI drops EC2 start/stop jobs in arm.yml and runs unit tests on GitHub-hosted ubuntu-22.04-arm, so merge/nightly no longer need secrets: inherit on the ARM workflow call.

EC2 runners for coverage and benchmarks add a Python step that normalizes github.actor into an AWS owner tag and passes aws-resource-tags (team=ai_and_search, owner) when starting runners via machulav/ec2-github-runner.

Debian CI replaces active Bullseye jobs with Debian Trixie (debian13.yml, gcc:14-trixie) in merge-queue and nightly flows, including pr-validation / notify-on-failure dependencies.

Tests only: FP16 override/batch tests in test_fp16.cpp and tiered SVS batch tests in test_svs_fp16.cpp scale vector data by 0.1f to keep squared L2 under FP16 max, use ASSERT_NEAR where SVE accumulation differs slightly—no production distance code changes.

^{Reviewed by Cursor Bugbot for commit cae7b11. Bugbot is set up for automated code reviews on this repo. Configure here.}

Additional merge-queue fix: backport the FP16 test-data scaling from #884 (MOD-13429), already used on newer release branches. This keeps squared L2 distances within FP16 range on the hosted ARM runner; no production distance kernels are changed.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 97.01%. Comparing base (cba8f74) to head (cae7b11).

Additional details and impacted files

@@           Coverage Diff           @@
##              8.2    #1040   +/-   ##
=======================================
  Coverage   97.01%   97.01%           
=======================================
  Files         126      126           
  Lines        7371     7371           
=======================================
  Hits         7151     7151           
  Misses        220      220           

☔ 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.