ci: install only the uv binary for v4 ingest (drop the workspace sync)

[connortsui20]

The v4 Postgres ingest runs uv run --no-project --with psycopg[binary] --with boto3 --with xxhash, which needs only the uv binary and an ephemeral env — never the workspace. But the "Install uv" step uses spiraldb/actions/setup-uv, which runs a full uv sync --all-extras --dev that rebuilds vortex-python via maturin → sccache → S3. On the extras=s3-cache bench runners that fails s3:GetObject AccessDenied under the just-assumed ingest role (which only has rds-db:connect).

It's continue-on-error, so CI never broke, and production confirmed the sync is dead weight — the ingest succeeded even when "Install uv" went red, because the binary installs before the sync and the ingest doesn't use the synced workspace. Switch the three v4 "Install uv" steps to astral-sh/setup-uv (the same v7.6.0 binary that action vendors) with no sync, which removes the wasteful workspace build and the sccache→S3 dependency entirely. The benchmark job's own non-v4 setup-uv is unchanged.

Follow-up to #8516, which fixed the same failure by reordering so the sync ran with good credentials; this drops the now-pointless sync instead, so step order no longer matters.

🤖 Generated with Claude Code

[codspeed-hq]

Merging this PR will improve performance by 18.2%

⚠️ Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data.

For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

⚡ 5 improved benchmarks
✅ 1576 untouched benchmarks

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	Simulation	chunked_varbinview_canonical_into[(1000, 10)]	191.6 µs	154.6 µs	+23.92%
⚡	Simulation	chunked_varbinview_opt_canonical_into[(1000, 10)]	206.7 µs	170.6 µs	+21.11%
⚡	Simulation	chunked_varbinview_opt_into_canonical[(1000, 10)]	221 µs	184.9 µs	+19.53%
⚡	Simulation	chunked_varbinview_canonical_into[(100, 100)]	258.8 µs	224.1 µs	+15.48%
⚡	Simulation	chunked_varbinview_opt_canonical_into[(100, 100)]	339.6 µs	304.9 µs	+11.37%

Tip

Curious why this is faster? Comment @codspeedbot explain why this is faster on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing ct/bench-v4-emitter-uv-binary (1b7139c) with develop (e0faa08)}