llm_directed: directed sex (breed offspring + select on verifier) — E10 in real weights

Adds the "directed sex" operator (E10) the moe regime-flip pointed to: don't
commit to one a-priori blend — breed a population of recombinant offspring
(specialists merged at Dirichlet-sampled weights), score each on a held-out
validation split with the verifier, and keep the fittest, reported on a fresh
test split. Two breeding objectives: best-overall and best-worst-family.
src/llm/directed.py + kind llm_directed, reusing the cached specialists.

Result — refinements pay off in proportion to how far the uniform soup is from
optimal:
- 0.5B (soup dilutes): directed selection beats soup on the bred objective —
  directed_overall 0.69 > soup 0.64; directed_balanced worst-family 0.37 > 0.26.
  Riders: single-objective selection trades off the other axis (overall-breed
  tanks lists to 0.17); a global blend still trails per-input routing (0.74).
- 7B (Imperial CX3, soup already composes to ceiling on near-saturated families,
  strings/arith 1.00): directed ~= soup (0.868 ~ 0.873, marginally below via a
  val/test overfit gap) — no fitter offspring to breed.

Through-line across all four LLM runs: "merge, don't average" and its refinements
(routing, directed selection) are weak-base / suboptimal-default phenomena — they
help at 0.5B and are inert at 7B. Honest limitation kept in the writeup: the 7B
families are near-saturated, which caps the headroom; a harder unsaturated
benchmark is the fair next test.

Also folds in the two llm_moe local manifest/config files missed in 8da0dac.
+3 directed unit tests (130 green). Results in results/llm_directed{,_hpc}/
(parquet gitignored).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

configs/llm/directed_hpc.yaml

@@ -0,0 +1,21 @@
+experiment: llm_directed_hpc
+kind: llm_directed
+seed: 1
+n_replicates: 1
+
+# Scaled directed-sex run for an L40S (48 GB). At 7B the uniform soup already composes (0.87, beating
+# every specialist), so the sharp question is whether SEARCHING the recombination-weight space +
+# selecting on the verifier can push past even that strong a-priori blend — i.e. does offspring
+# selection still add headroom once fusion is already strong? Reuses the cached 7B specialists.
+
+base_model: Qwen/Qwen2.5-7B-Instruct
+families: [lists, strings, arith]
+n_train: 800          # only if cached specialists are absent (fresh on the node)
+n_val: 100
+n_test: 200
+n_candidates: 24
+concentration: 0.5
+epochs: 3
+lora: {r: 16, alpha: 32}
+
+output: {dir: results/llm_directed_hpc}