sql: remove two private settings

[yuzefovich]

This commit removes two private settings that could disable planning of
multiple join readers (i.e would disable distribution) and disable
planning merge joiners (i.e. would simply prohibit the use of merge
joiners). It also retires two settings that have been removed after 20.1
was cut.

Release note: None

[cockroach-teamcity]

This change is 

[jordanlewis]

Is it valid to delete cluster settings? I know there's weird stuff if you load a binary on a database that has cluster settings that don't exist anymore. I run into it if I make a binary with a new cluster setting, run the binary, then go to current master without the cluster setting and rerun on the same data directory - there will be a log message "unknown cluster setting blah".

[yuzefovich]

I've just tried creating a store on master and disabling these two settings, then restarting cockroach with this PR, looked at all cluster settings (and the settings were removed), then restarted with the master sha and the settings were correctly set to false. So it looks like we no longer fail if settings are deleted. I also remember this behavior causing problems, but it appears as if we fixed it.

Couple of other settings that we could - but I'm less inclined to - remove are sql.defaults.zigzag_join.enabled and sql.distsql.interleaved_joins.enabled. Thoughts about these?

[petermattis]

@dt Informed me on another PR about the retired settings list.

[yuzefovich]

Oh, thanks Peter.

[asubiotto]

Reviewed 2 of 2 files at r1.
Reviewable status: complete! 1 of 0 LGTMs obtained

[yuzefovich]

TFTR!

bors r+

[craig]

Build succeeded

• GitHub CI (Cockroach)