fix(downloads): add the downloads package (was wrongly gitignored) + robust sanitizer

The .gitignore 'downloads/' pattern for the DOWNLOAD_ROOT bind mount also matched the
backend/app/downloads source package, so formats.py/storage.py/service.py never got
committed with M2. Anchor the ignore to '/downloads/' (repo root only) and add the package.

storage.sanitize() now handles arbitrarily messy titles (emoji, ZWJ, fullwidth, clickbait
punctuation): NFKC-normalize, drop emoji/symbol/control unicode categories, collapse
repeated punctuation, underscore-join words -> space-free paths. Accents (HU/DE) preserved,
no ASCII folding. Plex layout uses _-_ separators + Season_{year}.

backend/app/downloads/formats.py

@@ -0,0 +1,145 @@
+"""Format profiles → yt-dlp options, and the cache signature.
+
+A profile `spec` (stored on DownloadProfile / snapshotted on DownloadJob) is the single source
+of truth for what file gets produced. `format_sig` hashes exactly the fields that change the
+OUTPUT BYTES, so two users picking equivalent settings share one cached MediaAsset.
+`build_ydl_opts` turns a spec into a yt-dlp options dict (format selector + postprocessors).
+
+spec shape (all keys optional; defaults applied by `normalize`):
+    mode           "av" (audio+video) | "v" (video-only) | "a" (audio-only)
+    max_height     int | null (null = best)
+    container      "mp4" | "mkv" | "webm" | null (merge/remux target for av/v)
+    audio_format   "m4a" | "mp3" | "opus" | null (extract target for mode "a")
+    vcodec         "h264" | "vp9" | "av1" | null (preference, best-effort)
+    embed_subs     bool     embed_chapters bool     embed_thumbnail bool     sponsorblock bool
+"""
+import hashlib
+import json
+
+# Fields that affect the produced file — and therefore the cache identity.
+_SIG_KEYS = (
+    "mode",
+    "max_height",
+    "container",
+    "audio_format",
+    "vcodec",
+    "embed_subs",
+    "embed_chapters",
+    "embed_thumbnail",
+    "sponsorblock",
+)
+
+_DEFAULTS = {
+    "mode": "av",
+    "max_height": None,
+    "container": "mp4",
+    "audio_format": "m4a",
+    "vcodec": None,
+    "embed_subs": False,
+    "embed_chapters": True,
+    "embed_thumbnail": True,
+    "sponsorblock": False,
+}
+
+_SPONSORBLOCK_CATEGORIES = ["sponsor", "selfpromo", "interaction"]
+_VCODEC_SORT = {"h264": "vcodec:h264", "vp9": "vcodec:vp9", "av1": "vcodec:av01"}
+
+
+def normalize(spec: dict) -> dict:
+    """Fill defaults + coerce types so signature and yt-dlp opts are deterministic."""
+    s = dict(_DEFAULTS)
+    for k in _DEFAULTS:
+        if spec.get(k) is not None:
+            s[k] = spec[k]
+    if s["mode"] not in ("av", "v", "a"):
+        s["mode"] = "av"
+    if s["max_height"] is not None:
+        s["max_height"] = int(s["max_height"])
+    for b in ("embed_subs", "embed_chapters", "embed_thumbnail", "sponsorblock"):
+        s[b] = bool(s[b])
+    # Audio-only files can't embed subtitles/thumbnails as video; keep the flags meaningful.
+    if s["mode"] == "a":
+        s["embed_subs"] = False
+    return s
+
+
+def format_sig(spec: dict) -> str:
+    """Stable short hash of the output-affecting spec — the MediaAsset cache key component."""
+    s = normalize(spec)
+    payload = json.dumps({k: s[k] for k in _SIG_KEYS}, sort_keys=True, separators=(",", ":"))
+    return hashlib.sha256(payload.encode()).hexdigest()[:24]
+
+
+def target_ext(spec: dict) -> str:
+    """Best-effort final extension for the produced file (also used to name the staging output)."""
+    s = normalize(spec)
+    if s["mode"] == "a":
+        return s["audio_format"] or "m4a"
+    return s["container"] or "mp4"
+
+
+def build_ydl_opts(spec: dict, outtmpl: str, progress_hook) -> dict:
+    """Assemble yt-dlp options for a single download to `outtmpl` (a full path template)."""
+    s = normalize(spec)
+    h = s["max_height"]
+    opts: dict = {
+        "outtmpl": outtmpl,
+        "noplaylist": True,
+        "quiet": True,
+        "no_warnings": True,
+        "noprogress": True,
+        "writethumbnail": True,  # kept for the poster.jpg sidecar (and embed, if requested)
+        "progress_hooks": [progress_hook],
+        "postprocessors": [],
+    }
+
+    if s["mode"] == "a":
+        opts["format"] = "bestaudio/best"
+        opts["postprocessors"].append(
+            {
+                "key": "FFmpegExtractAudio",
+                "preferredcodec": s["audio_format"] or "m4a",
+                "preferredquality": "0",
+            }
+        )
+    else:
+        hcap = f"[height<=?{h}]" if h else ""
+        if s["mode"] == "v":
+            opts["format"] = f"bestvideo{hcap}/best{hcap}"
+        else:  # av
+            opts["format"] = f"bestvideo{hcap}+bestaudio/best{hcap}"
+        if s["container"]:
+            opts["merge_output_format"] = s["container"]
+        sort = []
+        if h:
+            sort.append(f"res:{h}")
+        if s["vcodec"] in _VCODEC_SORT:
+            sort.append(_VCODEC_SORT[s["vcodec"]])
+        if sort:
+            opts["format_sort"] = sort
+
+    # SponsorBlock first (marks segments), then strip them from the media + chapters.
+    if s["sponsorblock"]:
+        opts["postprocessors"].append(
+            {"key": "SponsorBlock", "categories": _SPONSORBLOCK_CATEGORIES, "when": "after_filter"}
+        )
+        opts["postprocessors"].append(
+            {"key": "ModifyChapters", "remove_sponsor_segments": _SPONSORBLOCK_CATEGORIES}
+        )
+    if s["embed_subs"] and s["mode"] != "a":
+        opts["writesubtitles"] = True
+        opts["subtitleslangs"] = ["en.*", "hu.*", "de.*"]
+        opts["postprocessors"].append({"key": "FFmpegEmbedSubtitle"})
+    if s["embed_chapters"]:
+        opts["postprocessors"].append(
+            {"key": "FFmpegMetadata", "add_chapters": True, "add_metadata": True}
+        )
+    # Normalize the written thumbnail to jpg so we always have a `<name>.jpg` poster sidecar.
+    # `already_have_thumbnail` keeps that file on disk after embedding (default deletes it).
+    opts["postprocessors"].append({"key": "FFmpegThumbnailsConvertor", "format": "jpg"})
+    if s["embed_thumbnail"]:
+        opts["postprocessors"].append(
+            {"key": "EmbedThumbnail", "already_have_thumbnail": True}
+        )
+
+    return opts