[Disk CI benchmarks] Use 1ES.Pool=diskann-github

[arrayka]

Switch .github/workflows/disk-benchmarks.yml and .github/workflows/disk-benchmarks-aa.yml to premium self-hosted VMs (L8s v3 Azure SKU with premium SSDs) to reduce noise and make performance gates more stable.

After this change, disk-based graph search throughput improved significantly - from 92 QPS to 596 QPS:

Before:
=======================================================================================================================================================
 L KNN      QPS  Mean Latency   95% Latency  99.9 Latency    IOs    IO (us)   CPU (us)   PQ Preprocess (us)  Mean Comps  Mean Hops  Cache Hit %  Recall
=======================================================================================================================================================
200 100     92.1     43391.1us       61602us      198334us  209.9  41693.3us   1657.8us               40.0us      5811.1      209.9         0.0%  91.506

After:
=======================================================================================================================================================
 L KNN      QPS  Mean Latency   95% Latency  99.9 Latency    IOs    IO (us)   CPU (us)   PQ Preprocess (us)  Mean Comps  Mean Hops  Cache Hit %  Recall
=======================================================================================================================================================
200 100    596.9      6687.5us        9923us       16727us  209.8   5406.7us   1248.3us               32.5us      5832.5      209.8         0.0%  91.428

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.60%. Comparing base (d516da1) to head (abeab51).
⚠️ Report is 13 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #869      +/-   ##
==========================================
- Coverage   90.60%   90.60%   -0.01%     
==========================================
  Files         461      459       -2     
  Lines       85559    85482      -77     
==========================================
- Hits        77525    77447      -78     
- Misses       8034     8035       +1     

Flag	Coverage Δ
miri	90.60% <ø> (-0.01%)	⬇️
unittests	90.56% <ø> (-0.01%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.
see 65 files with indirect coverage changes

🚀 New features to boost your workflow:

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

[Copilot]

Pull request overview

This PR moves disk benchmark workflows to the diskann-github 1ES self-hosted runner pool and adds local NVMe setup for benchmark workspace/data placement.

Changes:

• Switches disk benchmark and A/A benchmark jobs from ubuntu-latest to self-hosted 1ES runners.
• Adds NVMe formatting/mounting plus workspace symlinks in both benchmark workflows.
• Updates the shared disk benchmark setup action to install Rust if rustup is missing.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 4 comments.

File	Description
.github/workflows/disk-benchmarks.yml	Runs macro benchmarks on the 1ES self-hosted pool and stages workspace paths on NVMe.
.github/workflows/disk-benchmarks-aa.yml	Runs scheduled A/A benchmarks on the 1ES self-hosted pool and stages workspace paths on NVMe.
.github/actions/setup-disk-benchmark/action.yml	Adds conditional Rust/rustup installation before showing the toolchain and downloading datasets.

Comments suppressed due to low confidence (2)

.github/workflows/disk-benchmarks.yml:87

• These symlink creations are not idempotent. Self-hosted runner workspaces can persist between jobs, so a rerun or second matrix job on the same runner will fail here if diskann_rust or baseline already exists; remove/replace the links or use an idempotent workspace setup.

          ln -s /mnt/nvme/diskann_rust diskann_rust
          ln -s /mnt/nvme/baseline baseline

.github/workflows/disk-benchmarks-aa.yml:64

• These symlink creations are not idempotent. Self-hosted runner workspaces can persist between jobs, so a rerun or second matrix job on the same runner will fail here if diskann_rust or baseline already exists; remove/replace the links or use an idempotent workspace setup.

          ln -s /mnt/nvme/diskann_rust diskann_rust
          ln -s /mnt/nvme/baseline baseline

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[arkrishn94]

Just flagged one concern raised by Copilot. Want to make sure it is a non-issue?

Also, just out of curiosity - does the CI for this PR exercise these workflows?