Screenshots:
<img width="688" height="800" alt="image"
src="https://github.com/user-attachments/assets/9a234440-405a-428f-b5fe-a8969e94e359"
/>

<img width="1274" height="1842" alt="image"
src="https://github.com/user-attachments/assets/60e7b20d-45a9-465c-81fd-504ca8f4b931"
/>

## Summary

Frontend companion to #115337. Adds a "Download Images" action to the
snapshot header dropdown menu that fetches the zip from the new download
endpoint and triggers a browser download.

Also cleans up the `SnapshotImage` type and test fixtures:
- Removes `content_hash` from the frontend `SnapshotImage` interface —
it's an internal storage detail not needed by the UI
- Stops filtering `content_hash` from the metadata copy-to-clipboard
JSON (since the field is gone)
- Filters out `null` values from the metadata JSON to reduce noise
- Reorders dropdown menu items: Build Metadata → View Base Build →
Download Images → Rerun Status Checks → Delete

## Test plan
- [ ] Open a snapshot detail page and verify the dropdown menu shows the
new "Download Images" option
- [ ] Click "Download Images" and confirm the zip downloads with the
correct filename
- [ ] Verify the metadata copy button no longer includes `content_hash`
or null fields
- [ ] Confirm existing snapshot stories still render correctly without
`content_hash`

## Summary

- Adds a new streaming download endpoint
(`/preprodartifacts/snapshots/<id>/download/`) that bundles all snapshot
images into a zip file. Images are fetched concurrently in batches (32
workers, 200 per batch) and deduplicated by content hash before zipping.
The response is streamed as a `StreamingHttpResponse` — chunks are
yielded as each batch is zipped, so the client starts receiving data
immediately without buffering the entire archive in memory.
- This follows the same streaming download pattern already used by the
existing size build artifact download endpoint
(`project_preprod_artifact_download.py`), which also uses
`StreamingHttpResponse` for large file downloads.
- Uses a non-seekable `_DrainableBuffer` so `zipfile` writes data
descriptors (no seeking back to update headers), enabling true streaming
output.
- Replaces the fragile `**dict()` spread pattern in
`SnapshotImageResponse` construction with explicit field assignments
(`description`, `tags`). The old approach silently forwarded unknown
fields via `**kwargs` filtering, which was brittle and hard to follow.

## Test plan
- [ ] Hit the download endpoint for an existing snapshot and verify the
returned zip contains the expected image files
- [ ] Verify 404 is returned for non-existent snapshot IDs
- [ ] Verify 403 is returned when the feature flag is disabled
- [ ] Confirm that description and tags fields are still returned
correctly in the snapshot detail response

Screenshots:
<img width="688" height="800" alt="image"
src="https://github.com/user-attachments/assets/9a234440-405a-428f-b5fe-a8969e94e359"
/>

<img width="1274" height="1842" alt="image"
src="https://github.com/user-attachments/assets/60e7b20d-45a9-465c-81fd-504ca8f4b931"
/>

## Summary

Frontend companion to #115337. Adds a "Download Images" action to the
snapshot header dropdown menu that fetches the zip from the new download
endpoint and triggers a browser download.

Also cleans up the `SnapshotImage` type and test fixtures:
- Removes `content_hash` from the frontend `SnapshotImage` interface —
it's an internal storage detail not needed by the UI
- Stops filtering `content_hash` from the metadata copy-to-clipboard
JSON (since the field is gone)
- Filters out `null` values from the metadata JSON to reduce noise
- Reorders dropdown menu items: Build Metadata → View Base Build →
Download Images → Rerun Status Checks → Delete

## Test plan
- [ ] Open a snapshot detail page and verify the dropdown menu shows the
new "Download Images" option
- [ ] Click "Download Images" and confirm the zip downloads with the
correct filename
- [ ] Verify the metadata copy button no longer includes `content_hash`
or null fields
- [ ] Confirm existing snapshot stories still render correctly without
`content_hash`