Scan protocol traits

[joseph-isaacs]

Introduce composable scan planning through explicit IO and CPU state machines. Live Planner and Morsel implementations are Send and can move between threads with their owning run. Next<Input> constructs a live successor directly; constructors store inputs, while substantial work belongs in compute().

Requests are published exactly once through compute() returning NeedsIO(batch). state() reports Done, NeedsCompute, or Waiting; the driver tracks outstanding fetches, and waiting without one is a protocol error. IoSlot exposes is_waiting() instead of returning outstanding request batches.

Add the protocol under vortex_scan::planning and byte request/result contracts under vortex_io::request. Byte ranges carry alignment, IoSource supports completion polling with a waker, and IoService creates per-scope sessions. Include a simple blocking ReadAtIoSource adapter that declines optional hints. IO owner identities remain independent of scan scopes. Existing scan execution paths are unchanged.

Tests cover request bookkeeping, alignment, direct successor construction and constructor errors, and the reader adapter's registration, delivery, and cleanup behavior. No local builds, tests, lint, or formatting commands were run for this update.

First of three draft PRs: protocol and byte-source adapter → simple driver → footer opening/pruning. Data planners, IO priority/coalescing, and engine integration remain outside this PR. The later stacked PRs will need to adopt these updated contracts.

Stack order:

1. Scan protocol traits (this PR)
2. Simple scan planning driver
3. Composable footer opening and file pruning

[codspeed]

Merging this PR will not alter performance

⚠️ 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.

✅ 2105 untouched benchmarks
⏩ 503 skipped benchmarks¹

---

_{Comparing ji/scan-protocol-traits (5855fe8) with develop (8b7a257)}

Footnotes

1. 503 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[robert3005]

Let's go with this and tune as we learn more