docs: refresh guides with deploy-session learnings

fix(ui): make checkbox checked state visible

feat(ui): add advanced controls accordion — inference steps, cfg, infer method, seed, lm cot, schedule, metadata

fix(ui): restore desktop sidebar — don't apply min-width:0 to .ams-sidebar

fix(ui): stop wavesurfer controls + timestamps disappearing during re-render

fix(ui): stop layout 'dance' on mobile — cage audio waveform width

fix(deploy): redirect HF_MODULES_CACHE to /tmp — ~/.cache is read-only at runtime

fix(deploy): symlink hf preloads into fork's checkpoint dir + strip prompt tokens in lyrics

fix(deploy): skip cp -al mirror on zerogpu — exdev hardlink failure across filesystems

chore(debug): write startup prints to stderr (hf logs stream stderr more reliably)

chore(debug): add startup phase prints to diagnose hf spaces hang

feat(deploy): vendor ace-step as submodule, skip nano-vllm via sys.path injection

fix(deploy): switch to apple-silicon ace-step fork — no nano-vllm, looser gradio pin

fix(deploy): align gradio to ace-step's pin (6.2.0) — drop our own pin, let hf inject

fix(deploy): pin python_version 3.11 — hf spaces base bumped to 3.13 which breaks ace-step pin

fix(deploy): trim README short_description to fit HF Spaces 60-char cap

fix(lyrics): re-create mlx_lm generation_stream in worker thread

test(post): rewrite separate_stems test for demucs.apply.apply_model path

fix(deploy): align demucs warmup to runtime model name (htdemucs not htdemucs_ft)

feat(deploy): preload all hub models + warm demucs on spaces

fix(post): use lower-level demucs API + stringify stem paths for gr.Files

fix(lyrics): wrap mlx-lm generate in mx.stream(mx.gpu) for Gradio worker threads

fix(lora): force pipe load before unload_lora() when DiT is still cold

fix(deps): add hf_transfer for fast hub downloads

fix(deploy): pin gradio to 6.14 to match running version + claude.md note refresh

fix(deploy): per-mode duration estimator handles cover/extend/edit/lyrics signatures

fix(deploy): align m7 with claude.md — models/<org>/<repo> layout + gpu estimator

feat(deploy): m7 — hf spaces readme + cache bootstrap + zerogpu decorator

feat(ui): m6 polish — in-memory history sidebar + _safe_call error wrapper

feat(ui): expand tooltips.py + apply info= to cover/extend/edit/lyrics tabs

feat(post): wire separate stems / normalise / mp3 export buttons in output panel (m5 g2)

feat(post): add demucs stems, lufs normalisation, mp3 export

feat(ui): wire lyrics tab + cross-tab use-in-generate handoff (m4)

feat(lyrics): add qwen 2.5 7b lazy loader with mlx and transformers backends

feat(ui): wire cover/extend/edit tabs into the sidebar layout (m3 e3+e4)

feat(pipeline): wire cover/extend/edit task_types in studio.generate

feat(modes): add cover, extend, edit handlers with sub-mode dispatch

fix(lora): point manifest at the actually-published ACE-Step LoRAs

feat(ui): wire single-lora picker into generate tab (m2 d5)

feat(backend): apply lora stack before mode dispatch

feat(lora): add single-lora stack apply + disable

feat(lora): add preset registry with hf-download helper

feat(lora): add safetensors header sniff with ace-step module check

fix(ui): kill .styler bands + label icons + deepen pill contrast

fix(ui): full Brutalist Mono retheme — IBM Plex, padding crush, scoped selectors

fix(ui): wireframe-fidelity polish — neutral palette, mobile single-row, tight chrome

fix(ui): mobile sidebar — compact tab pills, kill dead vertical chrome

fix(ui): use gr.JSON for metadata output instead of gr.Code

fix(pipeline): pass save_dir so generate_music writes a usable file path

docs(plan): record checkpoint-layout findings from m1 smoke pass