Remove unused Puma cluster mode configuration

This removes the preload_app! directive and on_worker_boot block from
config/puma.rb. These cluster-mode features are not used since the
application runs in single-process mode (no workers configured).

VERIFICATION THAT CLUSTER MODE IS NOT USED:

1. No 'workers' setting in config/puma.rb - cluster mode requires
   explicitly setting 'workers N' where N > 0. Without this, Puma
   runs in single-process mode by default.

2. No WEB_CONCURRENCY environment variable - searched entire codebase,
   no references found. This is Heroku's standard variable for
   controlling worker count.

3. Procfile uses standard single-process command:
   'bundle exec puma -C config/puma.rb' with no worker-related flags.

4. No environment-specific Puma configs - only config/puma.rb exists,
   no production/staging overrides.

5. Environment configs (development/production/test.rb) have no Puma
   worker configuration.

6. The Puma 7.2.0 warning itself confirms the on_worker_boot block
   "will not execute in the current Puma configuration."

IMPACT:

- No behavior change - this code was never executing
- Removes warning introduced in Puma 7.x about unused callbacks
- Prepares for Puma 8.0 where on_worker_boot will be removed entirely
- If cluster mode is needed in future, these are standard patterns
  that can be easily restored

Related to #2447

Author: mroderick

config/puma.rb

@@ -34,12 +34,3 @@
 pidfile ENV["PIDFILE"] if ENV["PIDFILE"]
 worker_timeout 3600 if ENV.fetch("RAILS_ENV", "development") == "development"
 environment ENV.fetch("RAILS_ENV") { "development" }
-preload_app!
-# "worker" is the Puma term, not a background job.
-on_worker_boot do
-  ActiveSupport.on_load(:active_record) do
-    ActiveRecord::Base.establish_connection
-  end
-
-  SemanticLogger.reopen
-end