docs(fork-c/S0): correct two doc-accuracy lines (base-equivalence wording + grounded FFI count)

attempt-2 review fixes (docs-only; contract mechanics confirmed sound):
- README + S0.2 grep-clean: 'S0 HEAD == base' / 'S0 == base' were inaccurate
  (HEAD carries the docs/examples/tests diff). Reword to: production/compiler
  behavior is base-equivalent — zero src/ changes, single-author output
  byte-identical to base by construction — HEAD is a distinct commit, not base.
- S0.3 ledger: drop the stale '116-class corpus' FFI wording for the grounded
  live count (96 entry trees / 95 active markers), matching the S0.1 count note.

No partition / manifest / examples / harness change. Gate green:
zig build + zig build test (LSP sweep 574, no crash) + run_examples (540/0);
m3te ios-sim build via main binary exit 0.

docs/fork-c/S0.3-reuse-delete-ledger.md

@@ -22,7 +22,7 @@ is never merged (see `S0.2-…`).
 | **E-series selection rules** — own-wins / not-visible / ambiguity / direct-flat (the E1–E6a behaviors) | **resolver behavior + regression tests** (the baseline-green corpus is the mirror oracle) | S2 behavior; regressions locked S0 |
 | **CP rule** — body-author == layout-author | **keyed by `InstantiationId{template_decl, resolved_args}`** in the fact store | S4 |
 | **E6BR routed-signature cases** (the E6BR-1…4 behavioral cells) | **resolver-signature regressions** — the resolver walks every signature reference position; cases live in the resolver-target corpus, flip at S3.9 | S3.9 |
-| **FFI `foreign_class_map` consumers + 116-class corpus** | parallel `DeclId`s land at S1 (map still the consumer); foreign classes keyed by `DeclId` at S4; runtime names stay **payload strings on facts** | S1 → S4 |
+| **FFI `foreign_class_map` consumers + FFI corpus (96 entry trees / 95 active markers)** | parallel `DeclId`s land at S1 (map still the consumer); foreign classes keyed by `DeclId` at S4; runtime names stay **payload strings on facts** | S1 → S4 |
 
 ## B. DELETED / TRANSITIONAL — removed in S3/S6