fix(examples): repair datacenters SDK drift (US_GA_2 removed from enum)

[deanq]

Summary

04_scaling_performance/02_datacenters is broken against the current runpod_flash SDK: DataCenter.US_GA_2 was removed from the DataCenter enum, so gpu_worker.py raises AttributeError: US_GA_2 on import and flash dev fails to load it.

Changes

• Use DataCenter.US_KS_2 for the single-DC and multi-DC examples in gpu_worker.py.
• Prune data centers no longer in the enum (US-GA-2, US-MD-1, US-NC-1, EUR-IS-1) from the README table so it matches the current SDK.

Test plan

• flash dev now boots the example and discovers endpoints.
• multi_dc_inference returns COMPLETED (US-KS-2 + EU-RO-1); cpu routes OK.
• ruff format --check and ruff check pass (pre-commit quality-check green).

Note

Scope is datacenters-only. (07_benchmarking/01_dependency_benchmark is incomplete on main — its source lives on the unmerged tmp/07-benchmarking branch — so completing it is a separate merge decision, not part of this fix.)

[capy-ai]

Capy auto-review is paused for this organization because the usage-cycle auto-review limit has been reached. Increase the limit or turn it off in billing settings to resume automatic reviews.

[Copilot]

Pull request overview

Repairs two Flash examples that broke due to upstream runpod_flash SDK drift and restores missing benchmark example sources so flash dev can discover endpoints again.

Changes:

• Updated datacenter examples to use DataCenter.US_KS_2 (replacing removed US_GA_2) and aligned the README datacenter table with the SDK enum.
• Restored the 07_benchmarking/01_dependency_benchmark example (README + two benchmark endpoint files) so endpoints are discoverable again.
• Added CPU/system dependency benchmark implementations for numpy and ffmpeg, including version reporting and deterministic output hashing.

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 9 comments.

Show a summary per file

File	Description
07_benchmarking/01_dependency_benchmark/README.md	Restores documentation for dependency benchmarking and version-comparison workflow.
07_benchmarking/01_dependency_benchmark/numpy_benchmark.py	Restores a numpy benchmarking endpoint with pinned dependency version and result hashing.
07_benchmarking/01_dependency_benchmark/ffmpeg_benchmark.py	Restores an ffmpeg benchmarking endpoint using apt-installed system dependency and timing metrics.
04_scaling_performance/02_datacenters/README.md	Updates datacenter documentation/examples to match current SDK datacenter enum.
04_scaling_performance/02_datacenters/gpu_worker.py	Fixes broken datacenter enum usage by switching to a valid datacenter.

---

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

[Copilot]

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated no new comments.

[deanq]

Resolved the 9 Copilot review threads — all targeted 07_benchmarking/01_dependency_benchmark/, which was removed from this PR when it was narrowed to the datacenters-only fix. The PR now changes only 04_scaling_performance/02_datacenters/, and Copilot's re-review of that scope generated no comments.

Note: Copilot's benchmark feedback (use flash dev not flash run; prefix endpoint names with 07_01_) is valid and should be applied if/when that example lands via the tmp/07-benchmarking branch.

[deanq]

Closing as redundant. The datacenters SDK-drift fix (US_GA_2 -> US_KS_2 in 04_scaling_performance/02_datacenters/gpu_worker.py) already landed on main via a separate commit, so this branch adds nothing: git diff origin/main..origin/fix/example-sdk-drift is empty.

Verified 2026-07-06 by running the examples against published runpod-flash 1.18.0. Note the automated review's claim that this PR 'restores missing benchmark example sources' is inaccurate — the only diff was the datacenters file. The benchmark (07) and video_generator (06) sources remain unmerged on their own branches (tmp-07-benchmarking, ae-2222) and are out of scope here.

[deanq]

Correction to my previous comment: this PR is already MERGED (commit ecaf783, 2026-06-29) — that merge is what put US_KS_2 on main. My earlier wording ('landed via a separate commit') was wrong; the fix landed via this PR itself. Nothing to action here. Apologies for the noise.