perf(files): use bounded favorite searches in dashboard widget and activity helper

[joshtrichards]

• Resolves: #

Summary

Replace favorites ID lookup + per-node resolution with bounded file search queries in the Favorites widget and the activity helper.

What Changes

Previously these code paths:

• loaded favorite file IDs via tags
• resolved nodes one-by-one with getFirstNodeById()
• applied limits in PHP

This PR switches them to use the existing files search stack directly with:

• favorite = true
• explicit limit / offset
• explicit ordering (mtime DESC)

Why

This keeps the work bounded at the query layer and avoids repeated node lookups for small UI surfaces like the dashboard widget and favorites-based activity filtering.

Profiler traces on the dashboard widget showed the old implementation producing repeated filecache lookups per favorite, while the refactored implementation produces a single bounded query.

Notes

• Activity\Helper preserves the existing FAVORITE_LIMIT behavior by querying FAVORITE_LIMIT + 1 and still throwing "Too many favorites" if exceeded.
• This also makes favorite ordering deterministic. Previously the order from getFavorites() / getIdsForTag() was effectively unspecified.

Folow-up

• A future cleanup could add a shared files-layer helper/API for favorite-node searches so this query construction does not need to be repeated at call sites.
• There may be some opportunity to reduce queries elsewhere in the Activity look-ups used for the UI.

Checklist

• Code is properly formatted
• Sign-off message is added to all commits
• Tests (unit, integration, api and/or acceptance) are included
• Screenshots before/after for front-end changes
• Documentation (manuals or wiki) has been updated or is not required
• Backports requested where applicable (ex: critical bugfixes)
• Labels added where applicable (ex: bug/enhancement, 3. to review, feature component)
• Milestone added for target branch/version (ex: 32.x for stable32)

AI (if applicable)

• The content of this PR was partly or fully generated using AI