Redcarpet didn't like unordered lists deeply nested in an ordered one

I don't really like making these an ordered list since they're mutually
exclusive, but I think the structure is what's important.

[ci skip]

doc/migrate_ci_to_ce/README.md

@@ -111,14 +111,14 @@ each installation.
       # socket: /tmp/mysql.sock
     ```
 
-1. Depending on the values for `adapter`, you will have to use different
-   commands to perform the database dump.
+1. Depending on the values for `adapter`, you will have to use one of three
+   different commands to perform the database dump.
 
     **NOTE:** For any of the commands below, you'll need to substitute the
     values `IN_UPPERCASE` with the corresponding values from your **CI
     installation's** `config/database.yml` files above.
 
-      - If both your CI and CE (or EE) installations use **mysql2** as the `adapter`, use
+    1. If both your CI and CE (or EE) installations use **mysql2** as the `adapter`, use
        `mysqldump`:
 
         ```sh
@@ -129,7 +129,7 @@ each installation.
           ci_triggers ci_variables ci_web_hooks > gitlab_ci.sql
         ```
 
-      - If both your CI and CE (or EE) installations use **postgresql** as the
+    1. If both your CI and CE (or EE) installations use **postgresql** as the
        `adapter`, use `pg_dump`:
 
         ```sh
@@ -137,9 +137,9 @@ each installation.
           --data-only GITLAB_CI_DATABASE -t "ci_*" > gitlab_ci.sql
         ```
 
-      - If your CI installation uses **mysql2** as the `adapter` and your CE (or
-        EE) installation uses **postgresql**, use `mysqldump` to dump the database
-        and then convert it to PostgreSQL using [mysql-postgresql-converter]:
+    1. If your CI installation uses **mysql2** as the `adapter` and your CE (or
+       EE) installation uses **postgresql**, use `mysqldump` to dump the
+       database and then convert it to PostgreSQL using [mysql-postgresql-converter]:
 
         ```sh
         # Dump existing MySQL database first