(OPT) Traverse immutable byte slices without handwritten index scaffolding #780

Catalog ordering and Shift scheduling still spell out byte length, index, Option-get, increment, and loop control when reading an immutable Slice. The current source for-traversal admission accepts Vec only.

Add an explicit immutable byte-slice traversal shape with a Copy u8 item, lowering through the existing authenticated byte operations. Keep the held view and its exact root frozen across iteration; preserve left-to-right evaluation, early exits, failures, cleanup, and source identities. No new host ABI, ambient authority, owned-copy workaround, or resource-cap increase.

Implementation must join source verification, canonical source/graph projection, HIR resolution and independent validation, loans/cleanup, cache/workspace replay, interpreter, native, and Wasm. Include empty and nonempty slices, nested traversal and shadowing, exact/short builder budget and sticky refusal, early return/break/continue, failed byte access, forged/stale roots and forbidden owner mutation. Keep the manual-loop implementation as a differential reference.

Coordinate with scoped record-row loans (#775) where source-for lowering overlaps. Qualify in a large implementation batch with exact owning selectors, scoped Clippy, and only genuinely moved pin producers. No performance multiplier or token-saving claim until a matched live benchmark measures it.