feat(targets): re-add lemma/lemmy/lemrd full 6-variant slots (Phase 2 done)

The Phase 2 family rollout is complete — all three LetheanNetwork bases
now have their gguf (Q4_K_M / Q8_0 / BF16) and mlx (Q4 / 8bit / BF16)
quants uploaded, with paired lthn/* LEK-merged consumer repos.

Fleet is now at the documented target shape: 4 models × 6 variants =
24 paired A/B canons. Workers will pick up the new targets on their
next `lem-eval.sh maintain` cycle.

Co-Authored-By: Virgil <virgil@lethean.io>

targets.yaml

@@ -63,11 +63,116 @@ targets:
     base: hf.co/LetheanNetwork/lemer:BF16
     this: hf.co/lthn/lemer:BF16
 
-  # Lemma / Lemmy / Lemrd targets are temporarily parked here while the
-  # family rollout (Phase 2) is in progress — they'll be added back as
-  # (name, type, quant) triples mirroring lemer once the LetheanNetwork/
-  # bases have their gguf/mlx variants uploaded. Until then, workers
-  # only farm lemer to avoid fail-fast cascades on missing bases.
-  #
-  # See plans/claude/cladius/lem-benchmark-aggregator.md for the target
-  # state — the full fleet will be 4 models × 6 variants = 24 rows.
+  # Lemma — Gemma 4 E4B. Six fully paired quant slots.
+
+  - name: lemma
+    type: mlx
+    quant: Q4
+    base: LetheanNetwork/lemma-mlx
+    this: lthn/lemma-mlx
+
+  - name: lemma
+    type: mlx
+    quant: 8bit
+    base: LetheanNetwork/lemma-mlx-8bit
+    this: lthn/lemma-mlx-8bit
+
+  - name: lemma
+    type: mlx
+    quant: BF16
+    base: LetheanNetwork/lemma-mlx-bf16
+    this: lthn/lemma-mlx-bf16
+
+  - name: lemma
+    type: gguf
+    quant: Q4_K_M
+    base: hf.co/LetheanNetwork/lemma:Q4_K_M
+    this: hf.co/lthn/lemma:Q4_K_M
+
+  - name: lemma
+    type: gguf
+    quant: Q8_0
+    base: hf.co/LetheanNetwork/lemma:Q8_0
+    this: hf.co/lthn/lemma:Q8_0
+
+  - name: lemma
+    type: gguf
+    quant: BF16
+    base: hf.co/LetheanNetwork/lemma:BF16
+    this: hf.co/lthn/lemma:BF16
+
+  # Lemmy — Gemma 4 26B A4B MoE. Six fully paired quant slots.
+
+  - name: lemmy
+    type: mlx
+    quant: Q4
+    base: LetheanNetwork/lemmy-mlx
+    this: lthn/lemmy-mlx
+
+  - name: lemmy
+    type: mlx
+    quant: 8bit
+    base: LetheanNetwork/lemmy-mlx-8bit
+    this: lthn/lemmy-mlx-8bit
+
+  - name: lemmy
+    type: mlx
+    quant: BF16
+    base: LetheanNetwork/lemmy-mlx-bf16
+    this: lthn/lemmy-mlx-bf16
+
+  - name: lemmy
+    type: gguf
+    quant: Q4_K_M
+    base: hf.co/LetheanNetwork/lemmy:Q4_K_M
+    this: hf.co/lthn/lemmy:Q4_K_M
+
+  - name: lemmy
+    type: gguf
+    quant: Q8_0
+    base: hf.co/LetheanNetwork/lemmy:Q8_0
+    this: hf.co/lthn/lemmy:Q8_0
+
+  - name: lemmy
+    type: gguf
+    quant: BF16
+    base: hf.co/LetheanNetwork/lemmy:BF16
+    this: hf.co/lthn/lemmy:BF16
+
+  # Lemrd — Gemma 4 31B Dense. Six fully paired quant slots.
+
+  - name: lemrd
+    type: mlx
+    quant: Q4
+    base: LetheanNetwork/lemrd-mlx
+    this: lthn/lemrd-mlx
+
+  - name: lemrd
+    type: mlx
+    quant: 8bit
+    base: LetheanNetwork/lemrd-mlx-8bit
+    this: lthn/lemrd-mlx-8bit
+
+  - name: lemrd
+    type: mlx
+    quant: BF16
+    base: LetheanNetwork/lemrd-mlx-bf16
+    this: lthn/lemrd-mlx-bf16
+
+  - name: lemrd
+    type: gguf
+    quant: Q4_K_M
+    base: hf.co/LetheanNetwork/lemrd:Q4_K_M
+    this: hf.co/lthn/lemrd:Q4_K_M
+
+  - name: lemrd
+    type: gguf
+    quant: Q8_0
+    base: hf.co/LetheanNetwork/lemrd:Q8_0
+    this: hf.co/lthn/lemrd:Q8_0
+
+  - name: lemrd
+    type: gguf
+    quant: BF16
+    base: hf.co/LetheanNetwork/lemrd:BF16
+    this: hf.co/lthn/lemrd:BF16