Review pass: fix bugs, silence clippy, correct docs

Bugs:
- extract: pick the OUTERMOST archive suffix (sfx.first(), not last())
  so x.tar.gz uses the gz->tar path
- gwshop: truncate the error snippet by chars, not bytes (UTF-8 panic)
- sorter: py_stem_split now matches pathlib (0 < i < len-1) so a
  trailing dot is not treated as a suffix
- sorter::apply_plan: a per-file link/copy/mkdir failure is logged and
  skipped instead of aborting the whole apply (never-abort contract)
- orchestrate/state: add State::mark_logged and use it for every
  per-URL record so a state-save error cannot abort a run

Cleanups:
- remove dead is_archive_name and unused SourceInfo.manifest field
- mark kept-but-unused capabilities with allow(dead_code) + rationale
- let-chains / is_ok_and / is_multiple_of / replace([..]) / sort_by_key
  and other idiomatic simplifications across modules
- difflib: SequenceMatcher owns its char vecs (bit-identical parity kept)
- --only now enforces gofile|0807 via value_parser

Docs:
- SORTING.md: drop phantom sorter_plan.jsonl, os.link -> std::fs::hard_link,
  zipfile -> the zip crate
- README: add --since to the dlp synopsis, clarify --subs/--assume-done
- .gitignore: ignore catalogue/sorter/_logs/subscriptions run artifacts

Validation: cargo fmt --check clean, cargo clippy --all-targets 0 warnings,
cargo test 87 passed / 0 failed.

docs/SORTING.md

@@ -70,14 +70,14 @@ graph LR
    (`data/gwshop_catalog.json`) entries within the same faction; best
    sequence-match above `--threshold` (default 0.62) wins. The catalog
    provides canonical names, sub-factions and classes.
-4. **Plan** — default output: a dry-run table `source → target` plus
-   `sorter_plan.jsonl`. Nothing is touched.
+4. **Plan** — default output: a dry-run table `source → target` on stdout
+   (`--json` prints the same plan as JSON). Nothing is touched.
 5. **Apply** — create the target and **hardlink** the file there
-   (`os.link`); if the filesystems differ, fall back to a real copy with a
-   warning. Originals are never modified or deleted — that is what keeps the
-   *arr-style workflow possible (remux/seed/sort independently). Every
-   applied action is appended to `sorter_applied.jsonl` and re-runs skip
-   targets that already exist.
+   (`std::fs::hard_link`); if the filesystems differ, fall back to a real
+   copy with a warning. Originals are never modified or deleted — that is
+   what keeps the *arr-style workflow possible (remux/seed/sort
+   independently). Every applied action is appended to `sorter_applied.jsonl`
+   and re-runs skip targets that already exist.
 
 ## Bookkeeping location
 
@@ -95,8 +95,8 @@ confidence doc always lands next to the audit file.
   sees. Limitation: hardlinks need source and library on the same
   filesystem; the sorter detects this and copies instead (warned).
 - Zip contents **can** be read without extracting (that is how the sorter
-  peeks). Modifying a member in place is technically possible with
-  `zipfile` but is not useful here.
+  peeks). Modifying a member in place is technically possible with the `zip`
+  crate but is not useful here.
 - Endgame (later stage): unpack archives into the unit folder so the
   library is plain files, easier to sort per-model and to torrent. The
   unpack step will live in the sorter (`--unpack`, 7z binary) but stays off
@@ -142,7 +142,8 @@ already-present duplicates, 0 copied):
   (`noserverino` behavior): two paths to the same hardlinked file show
   different `st_ino`, and the values are stable across calls. To verify a
   link, compare `st_nlink` (and content/size) — never inodes. Server-side
-  links are real: a fresh `os.link` bumps nlink (2→3 in the checked case).
+  links are real: a fresh `std::fs::hard_link` bumps nlink (2→3 in the
+  checked case).
 - Filenames with raw cp437 bytes crash 7z direct-to-CIFS with exit 2
   ("Cannot open output file"); the extractor then extracts to a `/var/tmp`
   temp dir, repairs the names and moves the result over.