C4 Handoff: Extract ScdawgCore<U, V>

Goal

Factor ScdawgCore<U: CharUnit, V> parallel to SuffixAutomatonCore. The byte-keyed (Scdawg<V>, src/scdawg.rs) and char-keyed (ScdawgChar<V>, src/scdawg_char.rs) variants become thin wrappers.

Why this couldn't fit in one session

The audit estimated 3-4 weeks. Inspection:

• src/scdawg.rs — 1213 LOC, batch construction + IS-features (find, match_positions, count_substring) per Blumer 1987.
• src/scdawg_char.rs — 1045 LOC, parallel.

Differs from C3 (SuffixAutomatonCore) in that SCDAWG is batch- constructed; per-character on-line construction isn't supported. The generic factor would mirror C3's shape minus the per-character extend() state machine.

Pre-flight + step-by-step plan

Follow the C3 handoff structure (see c3-suffix-automaton-core-handoff.md) with these substitutions:

• SuffixAutomatonCore → ScdawgCore
• extend(u: U) step omitted; instead port compute_left_edges() and the batch from_terms(I) body.
• The Blumer "IS-features" (find / match_positions / count_substring) are inherent methods on the public type today; they need to become generic on ScdawgCore.

Expected LOC reduction

• scdawg.rs: ~1213 → ~50 LOC
• scdawg_char.rs: ~1045 → ~50 LOC
• New scdawg_core/: ~1200 LOC shared
• Net: ~900 LOC removed.

Risks

• The compute_left_edges() post-pass uses inner.write() and walks the whole node graph. Generification needs to keep this pass type-safe across U: CharUnit.
• The ScdawgNodeHandle<V> type today exposes ScdawgInner<V> internals. After generification this needs to become ScdawgCoreHandle<U, V> exposing ScdawgCoreInner<U, V>, with the same Dictionary / MappedDictionaryNode impls.