refactor: rename the application package to docsgpt

The backend import package is now docsgpt, the name it will carry on PyPI;
application was far too generic to install into anyone's site-packages.
git mv plus a mechanical rewrite of every import, dotted string and path
reference: 734 Python files, the compose files, Dockerfile, workflows, docs,
setup scripts, devcontainer, k8s manifests, vscode config, pytest and coverage
config, .gitignore. Behaviour is unchanged.

Kept for one release:
- A top-level application package whose meta-path finder resolves
  application.x.y to the already-imported docsgpt.x.y object, so old imports
  and entry points (celery -A application.app.celery,
  uvicorn application.asgi:asgi_app) keep working with a FutureWarning.
- Celery registers every application.* task name as an alias of its
  docsgpt.* task on start-up, so messages queued by the previous release still
  run. The redbeat key prefix moves to redbeat:docsgpt:v2: so schedule entries
  the previous release wrote are left unread instead of firing twice.

The backend image builds from the repository root (docker build -f
docsgpt/Dockerfile .) so it can ship the alias package; a root .dockerignore
allow-lists docsgpt/ and application/ and keeps caches, local data, .env
files, the sample index files and the Dockerfile out. Compose and the image
workflows point at the new context.

scripts/db/backfill.py

@@ -14,7 +14,7 @@ only knobs operators need. This script discovers what's available from
 the :data:`BACKFILLERS` registry and runs whichever tables were asked for.
 
 This script imports ``pymongo`` directly. ``pymongo`` is not part of the
-base ``application/requirements.txt`` post-migration — install it
+base ``docsgpt/requirements.txt`` post-migration — install it
 directly before running::
 
     pip install 'pymongo>=4.6'
@@ -49,13 +49,13 @@ sys.path.insert(0, str(Path(__file__).resolve().parents[2]))
 
 from sqlalchemy import Connection, text  # noqa: E402
 
-from application.core.settings import settings  # noqa: E402
-from application.storage.db.engine import get_engine  # noqa: E402
+from docsgpt.core.settings import settings  # noqa: E402
+from docsgpt.storage.db.engine import get_engine  # noqa: E402
 
 
 # The backfill tool is the one remaining consumer of MongoDB in this repo.
 # It reads from Mongo and writes to Postgres, so it keeps its own client
-# rather than going through the (now-deleted) ``application.core.mongo_db``
+# rather than going through the (now-deleted) ``docsgpt.core.mongo_db``
 # wrapper. The DB name is hard-coded to ``docsgpt`` — historically surfaced
 # as ``settings.MONGO_DB_NAME`` but that setting has been removed post-cutover.
 _MONGO_DB_NAME = "docsgpt"
@@ -889,7 +889,7 @@ def _rename_faiss_indexes(
         )
         return stats
 
-    from application.storage.storage_creator import StorageCreator
+    from docsgpt.storage.storage_creator import StorageCreator
 
     storage = StorageCreator.get_storage()
     storage_type = getattr(storage, "__class__", type(storage)).__name__
@@ -1302,14 +1302,14 @@ def _backfill_memories(
     Mongo memory docs don't carry the body inline — ``content`` lives on
     disk at ``doc["storage_path"]`` (e.g.
     ``inputs/local/memories/<oid>/memory.txt``) and is accessed through
-    :class:`application.storage.storage_creator.StorageCreator`. We read
+    :class:`docsgpt.storage.storage_creator.StorageCreator`. We read
     the file lazily here so the PG ``content`` column gets the actual
     memory text rather than an empty string. Missing/unreadable files are
     logged and fall back to an empty ``content`` so one bad row doesn't
     abort the whole batch. Import is lazy (matches ``_rename_faiss_indexes``)
     so ``storage`` / backend creds aren't required to import this module.
     """
-    from application.storage.storage_creator import StorageCreator
+    from docsgpt.storage.storage_creator import StorageCreator
 
     storage = StorageCreator.get_storage()
 

scripts/db/backfill_token_usage_model_id.py

@@ -1,7 +1,7 @@
 """Backfill ``token_usage.model_id`` for rows written before the column.
 
 New rows get ``model_id`` stamped at write time (see
-``application.llm.llm_creator`` / ``application.usage``). This script
+``docsgpt.llm.llm_creator`` / ``docsgpt.usage``). This script
 fills the historical NULLs by deriving the model from data we already
 trust, in priority order. A row is only ever filled by the
 highest-priority tier that matches it; tiers run in one transaction so
@@ -60,7 +60,7 @@ sys.path.insert(0, str(Path(__file__).resolve().parents[2]))
 
 from sqlalchemy import text  # noqa: E402
 
-from application.storage.db.engine import get_engine  # noqa: E402
+from docsgpt.storage.db.engine import get_engine  # noqa: E402
 
 
 # Tier 1: same request -> same model, primary (agent_stream) rows only.

scripts/db/backfill_tool_attempts_attribution.py

@@ -40,7 +40,7 @@ sys.path.insert(0, str(Path(__file__).resolve().parents[2]))
 
 from sqlalchemy import text  # noqa: E402
 
-from application.storage.db.engine import get_engine  # noqa: E402
+from docsgpt.storage.db.engine import get_engine  # noqa: E402
 
 
 # Tier 1: parent message → user, conversation → agent.

scripts/db/init_postgres.py

@@ -13,7 +13,7 @@ Intended use:
 
 This script is a thin wrapper around ``alembic upgrade head``. It exists
 separately so the same command is discoverable from the repo root without
-remembering the ``-c application/alembic.ini`` invocation.
+remembering the ``-c docsgpt/alembic.ini`` invocation.
 """
 
 from __future__ import annotations
@@ -25,7 +25,7 @@ from alembic import command
 from alembic.config import Config
 
 REPO_ROOT = Path(__file__).resolve().parents[2]
-ALEMBIC_INI = REPO_ROOT / "application" / "alembic.ini"
+ALEMBIC_INI = REPO_ROOT / "docsgpt" / "alembic.ini"
 
 
 def main() -> int:

scripts/db/migrate_model_ids.py

@@ -1,7 +1,7 @@
 """Rewrite stored model IDs across active config tables.
 
 Run this when a provider renames or deprecates a model ID. The catalog
-in ``application/core/models/<provider>.yaml`` is updated to the new ID,
+in ``docsgpt/core/models/<provider>.yaml`` is updated to the new ID,
 but existing agents and schedules still reference the old one and will
 fail on the next call. This script rewrites every active config row
 in-place inside a single transaction.
@@ -51,7 +51,7 @@ sys.path.insert(0, str(Path(__file__).resolve().parents[2]))
 
 from sqlalchemy import text  # noqa: E402
 
-from application.storage.db.session import db_session  # noqa: E402
+from docsgpt.storage.db.session import db_session  # noqa: E402
 
 
 # Built-in mapping reflects the 2026-05-25 Google preview -> GA swap.