Use uft8mb4 to support 4-byte unicode characters

[tiwillia]

Bug 1186681
https://bugzilla.redhat.com/show_bug.cgi?id=1186681
The utf8 character encoding in mysql supports unicode characters with 1-3 bytes. 4-byte unicode characters such as emojis require utf8mb4 encoding in mysql.

Fixes #6169

Since the encoding is set for a database in post_install, this update shouldn't interfere with existing mysql cartridges.

This pr also adds several environment variables. These are to set a default character set for client and server as well as to set a collation server and whether a client can specify a character set to use with its connection.

[tiwillia]

@mfojtik could you review? (you are listed as the mysql sme, lmk if I should ping someone else)

I'm unsure about the defaults in the my.cnf.erb file and whether these will impact existing mysql cartridges.

[mfojtik]

@tiwillia can you try to create app with current config (or tweak the setting back to current default), create some sample data and then switch it to utf8mb4?

I guess I will want to see prove that this change won't explode data in existing databases :-)

[tiwillia]

Since the database is created in post_install, the only change an existing application will see is additional configuration directives in ~/mysql/conf/my.cnf. I can verify that these changes don't explode or inhibit an existing database. I can still use the database, access it through mysql, and do a mysqldump.

However, I found an issue creating a rails application using the utf8mb4 type that might be an issue. Reviewing now, going to go ahead and [test][extended:cartridge]

[openshift-bot]

Online Test Results: FAILURE (https://ci.dev.openshift.redhat.com/jenkins/job/test_pull_requests/8846/)

[openshift-bot]

Evaluated for online test up to 010399e

[openshift-bot]

Online Action Required: Pull request cannot be automatically merged, please rebase your branch from latest HEAD and push again