docs: point users to --check and exact pack files

Troubleshooting, the FAQ and the download page sent people who only ran
the installer to verify.py, which reads the repository, not their BIOS
folder. Download links now name the exact asset of the recorded release,
the --dest example names the BIOS folder on the card, the slice layout
left with the last release that used it, and the shown verify output is
what the tool prints.

scripts/generate_readme.py

@@ -124,6 +124,34 @@ def release_totals(platform_name: str, record: dict) -> tuple[int | None, int |
     return pack["files"], pack["extracted_size"]
 
 
+def download_links(
+    platform_name: str, record: dict, repo: str, label: str, attrs: str = ""
+) -> str:
+    """Markdown links to the files of a platform's pack in the recorded release.
+
+    A pack published whole gets one link labelled *label*. A pack in parts
+    gets one link per part, since every part is needed. *repo* is the
+    repository URL, absolute or relative, and *attrs* a MkDocs attribute list
+    put after each link. Without a pack in the record the link falls back to
+    the latest release page.
+    """
+    pack = release_record.pack_for(platform_name, record)
+    tag = record.get("tag")
+    assets = pack.get("assets") if pack else None
+    if not tag or not assets:
+        return f"[{label}]({repo}/releases/latest){attrs}"
+    urls = [f"{repo}/releases/download/{tag}/{asset}" for asset in assets]
+    if len(urls) == 1:
+        return f"[{label}]({urls[0]}){attrs}"
+    parts = [
+        f"[{'P' if attrs else 'p'}art {number} of {len(urls)}]({url}){attrs}"
+        for number, url in enumerate(urls, 1)
+    ]
+    if attrs:
+        return " ".join(parts)
+    return f"{label}: " + ", ".join(parts)
+
+
 def extract_targets(platforms_dir: str) -> list[tuple[str, str]]:
     """(display name, folder) per registered platform, where its pack extracts.
 
@@ -182,7 +210,7 @@ def download_table(
             f"| {display} | {f'{files:,}' if files else '-'} |"
             f" {format_size(size) if size else '-'} |"
             f" {extract_paths.get(cov['platform'], '')} |"
-            f" [Download]({RELEASE_URL}) |"
+            f" {download_links(name, record, REPO_PATH, 'Download')} |"
         )
     return lines
 
@@ -219,7 +247,7 @@ def format_size(size: int) -> str:
 
 
 SITE_URL = "https://abdess.github.io/retrobios/"
-RELEASE_URL = "../../releases/latest"
+REPO_PATH = "../.."
 REPO = "Abdess/retrobios"
 
 
@@ -365,8 +393,9 @@ def generate_readme(db: dict, platforms_dir: str) -> str:
         "# Android (Termux, after pkg install python and termux-setup-storage)",
         "curl -fsSL https://raw.githubusercontent.com/Abdess/retrobios/main/install.sh | sh",
         "",
-        "# Handheld (SD card mounted on PC)",
-        "curl -fsSL https://raw.githubusercontent.com/Abdess/retrobios/main/install.sh | sh -s -- --platform retroarch --dest /path/to/sdcard",
+        "# SD card mounted on a PC: --dest is the BIOS folder on the card",
+        "# (RetroArch on Switch: retroarch/cores/system, on Android: RetroArch/system)",
+        "curl -fsSL https://raw.githubusercontent.com/Abdess/retrobios/main/install.sh | sh -s -- --platform retroarch --dest /path/to/sdcard/retroarch/cores/system",
         "```",
         "",
         "The script auto-detects your platform, downloads only missing files, and verifies checksums.",
@@ -381,16 +410,9 @@ def generate_readme(db: dict, platforms_dir: str) -> str:
         " The size is what the files occupy once extracted, and the ZIP itself"
         " downloads smaller.",
         "",
-        "A pack over 2 GB comes in several parts, and every part is needed."
-        " How to open them depends on their name:",
-        "",
-        "- `Pack.part1of2.zip`, `Pack.part2of2.zip`: each part is an ordinary"
-        " ZIP. Extract them all into the same folder.",
-        "- `Pack.zip.001`, `Pack.zip.002` (releases up to v2026.09.04): slices"
-        " of one ZIP, none of which opens on its own. Put them in one folder"
-        " and open the `.001` with 7-Zip or PeaZip, or join them first with"
-        " `cat Pack.zip.0* > Pack.zip` on Linux and macOS,"
-        " `cmd /c copy /b Pack.zip.001+Pack.zip.002 Pack.zip` on Windows.",
+        "A pack over 2 GB comes in several parts (`Pack.part1of2.zip`,"
+        " `Pack.part2of2.zip`), and every part is needed. Each part is an"
+        " ordinary ZIP: extract them all into the same folder.",
         "",
         "Every release ships `SHA256SUMS.txt` and a detached signature of it,"
         " checkable against `allowed_signers` in this repository:"
@@ -575,7 +597,8 @@ def generate_readme(db: dict, platforms_dir: str) -> str:
             "python scripts/generate_pack.py --list-emulators",
             "python scripts/generate_pack.py --list-systems",
             "",
-            "# Verify your BIOS collection",
+            "# Coverage of this repository's collection per platform, emulator or system",
+            "# (it reads database.json and bios/: for your own BIOS folder, run the installer with --check)",
             "python scripts/verify.py --all",
             "python scripts/verify.py --platform batocera",
             "python scripts/verify.py --emulator flycast",
@@ -734,13 +757,12 @@ pip install pyyaml "jsonschema[format-nongpl]" markdown
 python scripts/validate_pr.py --changed          # the BIOS files the branch adds
 python scripts/validate_schemas.py --source-only # profiles and platform files
 python scripts/profile_sync.py --emulator NAME --strict  # each profile the branch adds
+python scripts/refresh_data_dirs.py  # data directories the pack-building tests read
 python -m unittest discover -s tests -t .
 ```
 
 The full pipeline is not needed for a contribution: it rebuilds every pack
-(about 35 GB in `dist/`) and needs the data directories fetched first
-(`python scripts/refresh_data_dirs.py`), which is why maintainers run it
-before a release.
+in `dist/`, which is why maintainers run it before a release.
 
 ## PR validation