More tests for create_default_site

[Wojtek Ruszczewski]

There are some aspects of the sites.management.create_default_site handler that lack proper testing.

Tickets introducing or related to various pieces of it are: #7514, #15346, #15573, #16353, #16828, #17415 (the last one includes a test).

[Wojtek Ruszczewski]

The patch by ticket (some guessing about the problems):

• #7514 introduces a clear_cache statement, that's probably not needed any longer, because a pre_save signal clearing the cache was added in the meantime (the test the ticket refers to can be found ​here; removing the statement doesn't seem to cause any failures now);
• I believe the main issue in #15346 and #15573 was preventing the creation of more than one site;
• #16353 and #16828 are about a multi-database setup, just an allow_migrate condition is what is needed currently;
• #17415 has a nice test, but moving it to the new testcase makes it even better (by avoiding the SitesFrameworkTests.setUp), the test still fails with PostgreSQL if you remove the sequence resetting part in create_default_site.

Some cleanup of the patch is needed -- I suspect that removing the clear_cache statement and replacing 1 with SITE_ID should go as separate tickets.

[Tim Graham <timograham@…>]

In 1f98ec2e53e4636863396ab54f671f4546f9ba4c:

Fixed #23929 -- Added more tests for create_default_site.

Refs: #15346, #15573, #16353, #16828.

[Tim Graham <timograham@…>]

In 965a999ae5a03ca595f3842feacfbce5dc12f20a:

[1.7.x] Fixed #23929 -- Added more tests for create_default_site.

Backport of 1f98ec2e53e4636863396ab54f671f4546f9ba4c from master