[MM-60328] include proper search_path check to post-migrate

[isacikgoz]

Summary

After every migration, we should advise running migration-assist post-migrate command to determine the health of the DB in terms of schema and it's ownership.

Also make index creation optional. Will add documentation accordingly.

Ticket Link

https://mattermost.atlassian.net/browse/MM-60328

[galtom]

This "bug" affected me, and after hours of search I found the solution.
Whet the migration finished I tried to start Mattermost, but it is failed because SELECT CURRENT_SCHEMA(); SQL command returned nothing:

mattermost=> SELECT CURRENT_SCHEMA();
 current_schema
----------------

(1 row)

The SEARCH_PATH looked liked this:

mattermost=> SHOW SEARCH_PATH;
     search_path
---------------------
 """$user"", public"
(1 row)

But it is wrong. To make it work, I had to set by hand the it as postgres user:

ALTER ROLE mmuser SET search_path = "$user", public;

After that, this is what the SEARCH_PATH and CURRENT_SCHEMA() looked like:

mattermost=# SELECT CURRENT_SCHEMA();
 current_schema
----------------
 public
(1 row)

mattermost=# SHOW SEARCH_PATH;
   search_path
-----------------
 "$user", public
(1 row)

[agnivade]

From my understanding, https://mattermost.atlassian.net/browse/MM-60328 was about restoring the schema_search path in case the command failed to run successfully. But this PR just adds a post-migrate check correct?

Can we do something so that the search_path gets restored correctly in case of failure? If not, clearly mention that "command has failed, search_path could not be restored, please reset it manually" or something like that?