ActiveJob-aware wrapper that solves 'temp tables die between jobs'

[sncalvo]

Problem

The biggest footgun in the current API: if a user naively splits a staging workflow across multiple ActiveJob/Sidekiq jobs, each job checks out a different DB connection, and the temp staging table disappears between jobs. The README mentions the limitation but offers no fix beyond "stay on the same connection."

Proposal

A helper that auto-swaps to persistent staging (see the non-temp issue) when running inside an ActiveJob context:

class ImportUsersJob < ApplicationJob
  include StagingTable::JobHelper

  def perform(batch_id)
    staged_table(User, key: batch_id, temporary: false) do |staging|
      staging.insert(batch_for(batch_id))
    end
  end
end

class TransferUsersJob < ApplicationJob
  include StagingTable::JobHelper

  def perform(batch_id)
    staged_table(User, key: batch_id, temporary: false, create: false) do |staging|
      staging.transfer
    end
  end
end

• key: keeps the same persistent staging table across jobs
• The wrapper ensures drop_table happens exactly once (at the last job), or on failure/timeout
• Optional sweeper job for stranded tables older than TTL

Acceptance criteria

• Works with ActiveJob; Sidekiq-specific integration optional
• Stranded tables cleaned up by a configurable sweeper
• Docs: full worked example with job chain + failure handling