Group organisation usage by project and environment in one request

The usage dashboard can break usage down by request type and SDK, but not by project or environment. Those two need one request per key, and `GET /organisations/{id}/usage-data/` is throttled at five a minute per user, so any organisation with a handful of projects gets 429s.

The SDK dimension has no such problem because rows in `events_list` already carry `labels.user_agent`, so the frontend groups them client side off the one request the page has already made. Project and environment want the same shape.

**What we need.** One response that carries the per project and per environment split, rather than the caller asking once per key. Either:

- add project and environment to the labels on each row, so the existing client side grouping picks them up, or
- add `group_by=project|environment` returning one row per key

Whichever fits the analytics service better. The frontend only needs to be able to rank projects by usage without making N requests.

**Where things are**

- `api/app_analytics/views.py`, `get_usage_data_view`, carries `@throttle_classes([InfluxQueryThrottle])`
- `api/app_analytics/throttles.py`, scope `influx_query`, keyed on `user.pk`
- `api/app/settings/common.py`, `"influx_query": "5/min"`
- `api/app_analytics/serializers.py`, `UsageDataQuerySerializer` takes `project_id` and `environment_id` as filters today
- Both the Influx and Postgres paths in `api/app_analytics/analytics_db_service.py` would need it

**Why now.** #8186 shipped the breakdown with request type and SDK only. The project and environment dimensions are held back until this lands: a throttled scope reports nothing, and a breakdown that silently omits a project ranks the remaining ones wrongly.

Start in api/app_analytics/views.py at get_usage_data_view and trace the request through UsageDataQuerySerializer and analytics_db_service.py, including both Influx and Postgres paths. Review the existing request type and SDK breakdown from #8186, then verify the chosen response shape supports project and environment rankings from one request without per-key calls.

Written by the indexing model from the issue text.