erlang: lists keylist BIFs (keyfind/keymember/keydelete/keyreplace/keystore/keytake/keysort) (809/809)

Adds the tuple-keyed list family to lib/erlang/lists-ext.sx: act on
first match, key compare via == (er-equal?), non-tuples/short tuples
pass through. keysort/2 reuses the stable merge sort + full term
order. keytake/3 returns {value, Tuple, Rest} | false. All seven
registered through the er-register-builtin-bifs! wrapper so they
survive mid-run registry resets. lists_ext suite 17 -> 38.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

plans/erlang-on-sx.md

@@ -159,6 +159,8 @@ The Phase 9 opcodes are registered, tested, and bridged SX↔OCaml, but inert: n
 
 _Newest first._
 
+- **2026-06-30 stdlib hardening — `lists` keylists** — Added the keylist family to `lib/erlang/lists-ext.sx`: `keyfind/3`, `keymember/3`, `keydelete/3`, `keyreplace/4`, `keystore/4`, `keytake/3`, `keysort/2`. All operate on lists of tuples keyed on element N (1-indexed), act on the first match only, and pass through non-tuples / tuples shorter than N. Key comparison is `==` (`er-equal?`) per the stdlib; `keysort/2` reuses the stable `er-ext-msort` + `er-ext-lt?` from the sort commit, comparing extracted keys. `keytake/3` returns `{value, Tuple, Rest}` / `false`. Registered through the same `er-register-builtin-bifs!` wrapper so they survive registry resets. `lists_ext` suite 17→**38** (+21: hit/miss/first-match-only/short-tuple-skip across all seven, keysort by elem 1 and 2 + stability). Conformance **788 → 809/809**. Test-harness note: `element(2, T)` returns an integer (no `:name`), so those two cases compare the raw number via `erlang-eval-ast` rather than `er-lx-nm`. loops/erlang only.
+
 - **2026-06-30 stdlib hardening — `lists:sort/1,2` + `lists:usort/1`** — Roadmap is saturated within this loop's scope (every remaining `[ ]` is blocked: `httpc`/`sqlite` on absent host primitives, 10a/10c on out-of-scope `lib/compiler.sx`). Continued as forever-loop hardening by filling idiomatic-Erlang stdlib gaps. Added the `lists` sort family in a **new file `lib/erlang/lists-ext.sx`** (loaded after `runtime.sx`): stable merge sort over an SX-list bridge, registered via `er-register-pure-bif!`. `lists:sort/1` and `usort/1` use full Erlang term order; `sort/2` takes a `fun(A,B)->bool` comparator. **Two notable findings:** (1) the shared `er-lt?` (transpile.sx) only deep-compares numbers/atoms/strings and treats *any two tuples (or lists) as order-equal* — so `lists:sort` (and, latently, `min/2`/`max/2`) would not order compound terms. Fixed locally with a self-contained `er-ext-lt?` that compares tuples by arity-then-elementwise and lists elementwise (shorter proper prefix first), delegating cross-type cases to `er-lt?`. `er-lt?` itself left untouched (shared by the `<` operator; can't edit transpile.sx — see Blockers). (2) `tests/runtime.sx` resets the BIF registry mid-run via `er-register-builtin-bifs!`, which would wipe a one-shot registration; so `lists-ext.sx` **wraps** `er-register-builtin-bifs!` to re-add its BIFs on every rebuild. New `lists_ext` suite (17 tests: term order, dup-keeping, stability, descending comparator, usort dedup). Conformance **771 → 788/788** (12→13 suites). New-file workaround forced because every sx-tree write tool (incl. `sx_write_file`) raises yojson "Expected string, got null" in this worktree — authored via the `Write` fallback + `sx_validate`, the same pattern other loops use. loops/erlang only.
 
 - **2026-05-18 Phase 8 host-primitive BIFs wired (crypto / cid / file:list_dir)** — `loops/fed-prims` (merged at architecture `380bc69f`) delivered the platform primitives; wired the 3 previously-BLOCKED Phase 8 BIF groups in `lib/erlang/runtime.sx` as `er-register-pure-bif!`/`er-register-bif!` entries with term marshalling at the boundary. **`crypto:hash/2`** → `crypto-sha256`/`crypto-sha512`/`crypto-sha3-256`; atom `Type` dispatch, `er-source-to-string` for `Data`, host hex result → raw bytes via new `er-hexval`/`er-hex->bytes`, returns Erlang binary; bad type/arg → `error:badarg`. **`cid:from_bytes/1`** → `cid-from-bytes` with raw codec `0x55` + sha2-256 multihash assembled in SX (`[0x12,0x20]++digest`); **`cid:to_string/1`** → `cid-from-sx` of `er-format-value` (cbor-encode rejects `er-to-sx`-marshalled symbols; the canonical string form is total + deterministic). **`file:list_dir/1`** → `file-list-dir`, `{ok,[Binary]}` via `er-of-sx` / `{error,Reason}` reusing `er-classify-file-error`. Test gotcha caught + fixed: this Erlang port's binary parser only supports integer/var segments — `<<"abc">>` string-binary literals silently produce **empty** binaries, so the first-cut distinct-input tests compared two empty inputs and failed; rewrote ffi inputs to integer-segment binaries (`<<97,98,99>>`). ffi suite 14→**28** (3 BLOCKED negative-asserts flipped to positive+negative functional tests; `httpc`/`sqlite` kept as deferred unregistered-asserts per fed-prims handoff). Built `sx_server.exe` (dune, opam 5.2.0) at `380bc69f`; full conformance **729/729** (eval 385/385, vm 78/78, **ffi 28/28**, all process suites green). loops/erlang only — not merged, not pushed to architecture.