[release] Bump to version 65.0

- Investigate how historice lastfm imports work
- Fix LookupError on scrobble start/long-plays views
- Add Flexify workout importer
- Add workouts app with exercise catalog and workout routines
- When sharing a scrobble, share the log data too

PROJECT.org

@@ -18,7 +18,7 @@ tasks, Todoist tasks, web pages I've read and trails I've hiked has turned out
 to be sometimes cathartic and sometimes functional as I try to remember when I
 did a thing.
 
-* Backlog [5/29] :vrobbler:project:personal:
+* Backlog [0/24] :vrobbler:project:personal:
 ** TODO [#C] After transition to linux add curl_cffi as webpage scrapper again :webpages:metadata:
 :PROPERTIES:
 :ID:       d3cce1a7-d540-4d66-bf66-e75378e4eac7
@@ -557,109 +557,3 @@ Added a `workouts` Django app so we can scrobble gym sessions.
   and `tests/workouts_tests/` (unit conversion, form round-trips, importer,
   imperial POST) included.
 
-** DONE [#B] Investigate how historice lastfm imports work :importers:music:
-:PROPERTIES:
-:ID:       f91fdd53-9da7-4859-9322-f08b2c587061
-:END:
-*** Description
-
-Some time ago we added an ability for looking at the earliest LastFM import for
-a given user and to kick off a series of monthly imports of past lastfm
-scrobbles until there are no more. I tried running this a while ago and it
-didn't work, and I wonder where we left it off.
-
-*** Investigation findings
-
-The historical import mechanism (`dispatch_historical_imports` in
-`vrobbler/apps/scrobbles/importers/lastfm.py`) was added in commit `4e1c3ffb`
-and ran successfully on 2026-05-24, importing 1,450 scrobbles covering
-Feb-Apr 2023 (imports 1, 4, 5) — the user's full Last.fm history back to
-registration.
-
-Two real bugs were found and fixed:
-
-- Resume cursor: the dispatch loop found the import with the earliest
-  `processed_finished` instead of the import covering the oldest scrobbles,
-  so re-running would re-dispatch already-imported months. It now walks back
-  from the earliest scrobble across all completed imports.
-- Dedup: `import_from_lastfm` filtered existing scrobbles by `created`
-  (row-insert time) against the Last.fm play timestamp ± 20s, which never
-  matches for historical data and caused duplicate scrobbles on re-import.
-  It now compares `timestamp`. Also fixed a `tzinfo.name` crash for
-  `ZoneInfo` timezones.
-
-Tests added in `tests/scrobbles_tests/test_lastfm.py` (5 tests) covering
-dedup-by-timestamp, creation, and dispatch resume/no-resume behavior.
-
-** DONE [#B] Fix LookupError on scrobble start/long-plays views :bug:scrobbles:videos:
-:PROPERTIES:
-:ID:       868e2a91-c974-4656-a8ae-b90a1f94ae74
-:END:
-
-- `PLAY_AGAIN_MEDIA["channels"]` and `LONG_PLAY_MEDIA["papers"]` keyed models
-  by app labels that don't exist (`Channel` lives in `videos`, `Paper` in
-  `books`), so `scrobble_start`, `scrobble_longplay_finish`, and the long-plays
-  views raised `LookupError: No installed app with label 'channels'` for every
-  request.
-- `scrobble_start` now special-cases `channels` to the `videos` app and skips
-  unknown app labels; a shared `get_long_play_media_model()` helper in
-  `scrobbles/utils.py` maps the `papers` label to `books` for the long-play
-  lookups.
-- Also fixed `Channel.scrobble_for_user` which still used the pre-refactor
-  `Scrobble.create_or_update(media_object=...)` signature (would raise
-  `TypeError` once the lookup reached it).
-- Fixed in commit `fbb1725d`; audited all other `apps.get_model` callers of
-  the media maps (`AUTO_FINISH_MEDIA` uses class-name values only, harmless).
-  Regression tests: channel/video start, long-plays view.
-** DONE [#B] Add Flexify workout importer :workouts:importer:
-:PROPERTIES:
-:ID:       96c9fbf5-a0d5-4f09-a72c-bb491b7ece18
-:END:
-
-*** Description
-
-Import gym sessions and routines from a Flexify backup.
-
-- `import_flexify()` in `workouts/importer.py` reads a Flexify sqlite backup
-  (`data/flexify-example.sqlite`) plus an optional plans CSV
-  (`data/flexify-plans-example.csv`). Plans become `WorkoutRoutine`s with
-  `WorkoutRoutineExercise` planned-exercise rows; logged `gym_sets` are
-  clustered into sessions (gap > 60 min or plan change) and converted into
-  workout scrobbles.
-- A curated `FLEXIFY_EXERCISE_ALIASES` map resolves Flexify exercise names to
-  the wrkout catalog, falling back to `find_or_create` for unmatched names.
-- `python manage.py import_flexify --sqlite ... --csv ... --user ...` command.
-- `scan_webdav_for_flexify()` watches `var/workouts/` on WebDAV (via the
-  existing import_from_webdav sweep), pairing a plans CSV with the sqlite
-  backup, and queues `process_workout_import` celery tasks.
-- `WorkoutImport` model (`BaseFileImportMixin`, new `file_hash` field for
-  change detection), admin, DRF `workout-imports` viewset, planned-exercises
-  block on the routine detail template, and MCP routine dict update.
-** DONE [#B] Add workouts app with exercise catalog and workout routines :workouts:feature:
-** DONE [#A] When sharing a scrobble, share the log data too :bug:scrobbles:sharing:
-:PROPERTIES:
-:ID:       46de9f0e-2727-63f4-0845-e714dbbe4a45
-:END:
-
-- Note taken on [2026-08-07 Fri 15:02]
-  
-  An example is here: https://life.lab.unbl.ink/shared/I3fjpLS/
-
-  The agent session has no data when shared, becuase it's all stored in logdata.
-
-  There may be some logdata that's sensitive, look out for that and maybe put it
-  in a SENSITIVE_LOGDATA_KEYS dictionary in settings and exclude that from being
-  shown.
-
-- Note taken on [2026-08-07 Fri]
-  
-  Fixed. The share template (`scrobble_share.html`) now renders logdata for
-  AgentSession (the conversation) and Workout scrobbles, matching the detail
-  template. The shared view renders a non-interactive conversation (no follow-up
-  form, retry buttons, or htmx polling for anonymous viewers).
-
-  Added `SENSITIVE_LOGDATA_KEYS` in `vrobbler/settings.py` keyed by media type.
-  `Scrobble.public_log` and `Scrobble.public_logdata` strip those keys for
-  shared/public rendering; owner views keep the full `log`/`logdata`. Defaults
-  cover body-composition metrics on Task scrobbles and `bodyweight_kg` on
-  Workout scrobbles.