sql: refactor SynthesizeRegionConfig usages

[otan]

Use the option method instead to decide to read from the cache.

Release note: None

[cockroach-teamcity]

This change is 

[arulajmani]

This may break the following (contrived) scenario:

• Client 1: ALTER DATABASE ... DROP REGION "us-east"
• Client 2: CREATE TABLE t() LOCALITY REGIONAL BY ROW
• Client 3, which (say) has version v-1 of the type descriptor, validate_multi_region_zone_configs() <--- Fails because it expects t to have a partition for "us-east"

This is because on version v-1 "us-east" is public, so we'll validate for it. Granted, this is quite contrived, but maybe one round-trip is a fair enough price to pay here for our sanity? Wdyt?

Reviewable status: complete! 0 of 0 LGTMs obtained

[otan]

I would think client 3 in this case would also have v1 of the zone config.

[arulajmani]

Maybe I'm missing something, but aren't we reading zone configs from the system.zone table?

Also, independent of round trip/no round trip, I like the way this PR cleans up the code so we should get it in in some form.

Reviewable status: complete! 0 of 0 LGTMs obtained

[otan]

Would've thought leasing would also mean the reads would be at an earlier timestamp when reading system.zones, but maybe I'm jumping too many hurdles. I'll revert the round-trip bit for now, but keep the refactor for SynthesizeRegionConfig

[arulajmani]

Reviewable status: complete! 1 of 0 LGTMs obtained (waiting on @otan)

---

pkg/sql/region_util.go, line 889 at r1 (raw file):

// CurrentDatabaseRegionConfig is part of the tree.EvalDatabase interface.
// CurrentDatabaseRegionConfig uses the cache to synthesize the RegionConfig
// and as such is intended for DML use. It returns an nil

nit: s/returns an nil/returns nil/g

[otan]

bors r=arulajmani

[craig]

🕐 Waiting for PR status (Github check) to be set, probably by CI. Bors will automatically try to run when all required PR statuses are set.

[craig]

Build succeeded:

• GitHub CI (Cockroach)