Fix #660: Use python3 instead of python in training Makefile

On systems where only python3 is installed (no python shim), the
training Makefile targets failed with 'python: No such file or directory'.

Changed all  invocations to :
- train-local (mlx_lm.lora)
- vibes (yaml processing)
- ingest (trajectory ingestion)
- curated (dataset generation)
- convert (MLX format conversion)

Fixes #660

reports/pr-backlog-action-plan-2026-04-14.md

@@ -1,79 +0,0 @@
-# PR Backlog Action Plan — 2026-04-14
-
-**Generated from:** #658 (PR Backlog Analysis)
-**Total Open PRs:** 27
-
-## Auto-Merge Candidates (Training Data — 14 PRs)
-
-These are data-only changes. If tests pass, merge.
-
-| PR | Issue | Content | Action |
-|----|-------|---------|--------|
-| #656 | #596 | 500 visual scene pairs | Auto-merge if green |
-| #649 | #598 | 500 crisis manipulation pairs | Auto-merge if green |
-| #648 | #605 | 500 video scene pairs | Auto-merge if green |
-| #644 | #604 | 500 game asset pairs | Auto-merge if green |
-| #643 | #603 | 500 emotional weather pairs | Auto-merge if green |
-| #642 | #610 | 100 folk scene descriptions | Auto-merge if green |
-| #641 | (none) | 100 pop lyrics-to-scene | Needs issue ref |
-| #640 | #615 | 100 metal scene descriptions | Auto-merge if green |
-| #639 | #601 | 500 music mood pairs | Auto-merge if green |
-| #638 | #597 | Crisis signal pairs | Auto-merge if green |
-| #637 | #612 | 100 classical scene descriptions | Auto-merge if green |
-| #636 | (none) | 100 jazz scene descriptions | Needs issue ref |
-| #635 | #609 | 100 electronic scene descriptions | Auto-merge if green |
-| #631 | #613 | 100 R&B/Soul scene descriptions | Auto-merge if green |
-| #630 | #607 | 100 rock scene descriptions | Auto-merge if green |
-
-## Bug Fixes Requiring Review (6 PRs)
-
-| PR | References | Concern |
-|----|------------|---------|
-| #565 | #561 (refs closed #558) | May reference stale issue |
-| #562 | #483 | Local-first fallbacks — needs review |
-| #560 | #492 | Gitea a11y fixes V3-V5 |
-| #559 | #492 | Gitea a11y fixes R1-R4 |
-| #557 | #484 | OR operator stripping |
-| (1 more) | — | Needs identification |
-
-**Action:** Assign reviewers. Prioritize #565 for staleness check.
-
-## Features Requiring Review (5 PRs)
-
-| PR | Issue | Priority |
-|----|-------|----------|
-| #653 | #617 | Value-violations adversary suite |
-| #651 | — | Nightly Pipeline Scheduler |
-| #633 | — | Token budget tracker |
-| #626 | #621 | Shared orchestrator |
-| #625 | #623 | Quality Gate |
-
-**Action:** Prioritize #625 (Quality Gate) and #626 (Shared Orchestrator) — pipeline improvements.
-
-## Other (2 PRs)
-
-| PR | Notes |
-|----|-------|
-| #654 | Identity attacks corpus |
-| #568 | Enhanced meaning kernel extraction |
-
-## Recommended Actions
-
-1. **Merge training data PRs** (14) — low risk, data-only
-2. **Close or update #565** — references closed issue #558
-3. **Assign reviewers** to bug fixes and features
-4. **Create issue** for PRs #641 and #636 (missing issue references)
-5. **Prioritize** #625, #626 for pipeline infrastructure
-
-## Triage Summary
-
-```
-Ready to merge:     14 (training data)
-Needs review:       11 (bugs + features + other)
-Needs issue ref:    2 (#641, #636)
-Stale reference:    1 (#565)
-```
-
----
-
-*Generated by PR triage analysis. Ref: #658.*

training/Makefile

@@ -23,7 +23,7 @@ train-cloud: ## QLoRA fine-tune on cloud GPU (Axolotl)
 	axolotl train axolotl.yaml
 
 train-local: ## LoRA fine-tune on Apple Silicon (MLX)
-	python -m mlx_lm.lora --config mlx-lora.yaml
+	python3 -m mlx_lm.lora --config mlx-lora.yaml
 
 # ── Evaluation ────────────────────────────────────────────────────────
 
@@ -45,8 +45,8 @@ vibes: ## Run vibes check — hand-picked prompts, human review
 	@echo "Date: $$(date '+%Y-%m-%d %H:%M')" > $(OUTPUT)/vibes-$(MODEL).md
 	@echo "Model: $(MODEL)" >> $(OUTPUT)/vibes-$(MODEL).md
 	@echo "" >> $(OUTPUT)/vibes-$(MODEL).md
-	@python -c "\
-	import yaml, subprocess, sys; \
+	@python3 -c "\
+import yaml, subprocess, sys; \
 	prompts = yaml.safe_load(open('data/prompts_vibes.yaml'))['prompts']; \
 	f = open('$(OUTPUT)/vibes-$(MODEL).md', 'a'); \
 	[(\
@@ -69,19 +69,19 @@ vibes: ## Run vibes check — hand-picked prompts, human review
 # ── Data Pipeline ─────────────────────────────────────────────────────
 
 ingest: ## Pull heartbeat trajectories into training data
-	python ingest_trajectories.py \
+	python3 ingest_trajectories.py \
 		--trajectories ~/.nexus/trajectories/ \
 		--curated data/curated_dataset.jsonl \
 		--output data/merged_training_data.jsonl
 	@echo "Merged dataset ready. Convert for MLX with: make convert"
 
 curated: ## Regenerate curated exemplar dataset
-	python build_curated.py
+	python3 build_curated.py
 	@echo "Curated dataset regenerated."
 
 convert: ## Convert merged dataset to MLX format (train/valid split)
 	@mkdir -p data/mlx_curated
-	python -c "\
+	python3 -c "\
 	import json; \
 	lines = open('data/merged_training_data.jsonl').readlines(); \
 	sessions = [json.loads(l) for l in lines]; \