PERF: migrate normalized_emails in a migrations (discourse#15166)

Old OnceOff job could perform pretty slowly on sites with millions of emails

New implementation operates in batches in a migration, minimizing locking.

db/post_migrate/20211201221028_migrate_email_to_normalized_email.rb

@@ -0,0 +1,50 @@
+# frozen_string_literal: true
+
+class MigrateEmailToNormalizedEmail < ActiveRecord::Migration[6.1]
+
+  # minimize locking on user_email table
+  disable_ddl_transaction!
+
+  def up
+
+    min, max = DB.query_single "SELECT MIN(id), MAX(id) FROM user_emails"
+    # scaling is needed to compensate for "holes" where records were deleted
+    # and pathological cases where for some reason id 100_000_000 and 0 exist
+
+    # avoid doing any work on empty dbs
+    return if min.nil?
+
+    bounds = DB.query_single <<~SQL
+      SELECT t.id
+      FROM (
+        SELECT *, row_number() OVER(ORDER BY id ASC) AS row
+        FROM user_emails
+      ) t
+      WHERE t.row % 100000 = 0
+    SQL
+
+    # subtle but loop does < not <=
+    # includes low, excludes high
+    bounds << (max + 1)
+
+    low_id = min
+    bounds.each do |high_id|
+
+      # using execute cause MiniSQL is not logging at the moment
+      # to_i is not needed, but specified so it is explicit there is no SQL injection
+      execute <<~SQL
+        UPDATE user_emails
+        SET normalized_email = REPLACE(REGEXP_REPLACE(email,'([+@].*)',''),'.','') || REGEXP_REPLACE(email, '[^@]*', '')
+        WHERE (normalized_email IS NULL OR normalized_email <> (REPLACE(REGEXP_REPLACE(email,'([+@].*)',''),'.','') || REGEXP_REPLACE(email, '[^@]*', '')))
+          AND (id >= #{low_id.to_i} AND id < #{high_id.to_i})
+      SQL
+
+      low_id = high_id
+    end
+
+  end
+
+  def down
+    execute "UPDATE user_emails SET normalized_email = null"
+  end
+end