Add 'Reset Data' endpoint and UI to purge all projects/analyses/datapoints

[anchapin]

Problem

Users currently have no way to start with a clean data state through the web UI or API. The only options are:

1. Individual deletion via CRUD (tedious for bulk cleanup)
2. rake db:purge (command-line only, no UI)
3. nuke.sh scripts (tear down the entire Docker deployment — overkill)
4. Admin restore (replaces DB with a backup, but not a "clean slate")

Current State

What exists:

• Individual project/analysis/datapoint delete via DELETE /projects/:id, DELETE /analyses/:id, DELETE /data_points/:id
• Cascading dependent: :destroy on Project→Analyses→DataPoints (with filesystem cleanup via background jobs)
• Admin page (/admin) with: backup, restore, prune dead Resque workers, requeue failed jobs
• Mongoid's built-in rake db:purge (drops all collections)

What's missing:

• No DELETE /admin/projects or POST /admin/reset endpoint
• No "Delete All" or "Reset" button in the admin UI
• No rake task for selective cleanup (e.g., purge completed analyses only)

Proposed Implementation

API

Add POST /admin/reset endpoint that:

1. Drops all collections in the database (equivalent to db:purge)
2. Cleans up filesystem artifacts under /mnt/openstudio/server/assets/analyses/
3. Returns confirmation of what was deleted

Optional: Add DELETE /admin/projects for project-only purge (less destructive).

Web UI

Add a "Reset All Data" button to server/app/views/admin/index.html.erb:

• Confirmation dialog: "This will permanently delete ALL projects, analyses, and datapoints. Are you sure?"
• Calls POST /admin/reset via form submission
• Shows success/failure message

Rake Task

Add rake data:reset for command-line use (mirrors the API endpoint).

Files to Modify

• server/app/controllers/admin_controller.rb (add reset action)
• server/config/routes.rb (add route)
• server/app/views/admin/index.html.erb (add reset button + confirmation)
• server/lib/tasks/ (add rake task)

Safety Considerations

• Should require admin authentication (already in place for admin controller)
• Should log the reset action
• Consider adding a "soft reset" that only removes completed/failed analyses, not running ones
• Filesystem cleanup should be async (background job) to avoid blocking the request