feat(post-bench): 3-branch decision pipeline ready before bench fires

rename: drop '-lora-' segment from all model names + capitalize v1.5 size

feat(v1.5): 32B SFT config + 3-way benchmark (v1 vs base32B vs v1.5)

feat(v2): upgrade base model 32B → 72B (Qwen2.5-72B-Instruct)

fix(data-feed): repair dataset feeding gaps

feat(rounds-7-12-clean-no-tokens): SHARD + ZeroGPU + personae + monitor

feat(v2-round5): sustainability loops + 2026 techniques

feat(v2): ambitious targets + 5 free-technique implementations

feat(v2): Phase A complete build infrastructure ready to execute