release-23.1.20-rc: opt: add GenerateTrigramSimilarityInvertedIndexScans rule

[mgartner]

Backport 5/5 commits from #121973.

/cc @cockroachdb/release

---

sql: add optimizer_use_trigram_similarity_optimization setting

This commit adds a new session setting called
optimizer_use_trigram_similarity_optimization. It currently has no
effect, but in future commits it will allow control over a new
optimization for queries with trigram similarity filters.

Release note: None

sql: disable inverted scans when similarity optimization is enabled

The GenerateInvertedIndexScans rule is now disabled for similarity
filters on trigram indexes when the
optimizer_use_text_similarity_optimization session setting is enabled.
In a future commit this setting will enable a new rule that makes
GenerateInvertedIndexScans obsolete. There is no reason to trigger
both rules.

Release note: None

opt: add GenerateTrigramSimilarityInvertedIndexScans rule

The GenerateTrigramSimilarityInvertedIndexScans exploration rule has
been added, which index-accelerates trigram similarity filters. See the
comment above the newly added custom function for more details. In
future commits, this rule will be optimized further.

The tests for this rule break from convention: they are added to a new
file rather than to the select test file corresponding to the location
of the rule in select.opt. This proposed new method of organization is
motivated by the huge growth of the select file over the years. If
there is agreement on this, we can incrementally reorganize all existing
tests to match.

Release note (performance improvement): More efficient query plans are
now generated for queries with text similarity filters, e.g.,
text_col % 'foobar'. These plans are generated if the
optimizer_use_trigram_similarity_optimization session setting is
enabled. It is disabled by default.

opt: reduce scanned trigrams for similarity filters

The GenerateSimilarityInvertedIndexScans exploration rule now
generates plans that scan fewer trigrams. See the added code comments
for more details.

Release note: None

opt: consolidate logic for constraining inverted index prefix columns

The logic for building constraints on prefix columns of multi-column
inverted indexes has been refactored and consolidated.

Fixes #112675

Release note: None

---

Release justification: Performance improvements gated behind a
session setting.

[blathers-crl]

Thanks for opening a backport.

Please check the backport criteria before merging:

• Backports should only be created for serious
  issues or test-only changes.
• Backports should not break backwards-compatibility.
• Backports should change as little code as possible.
• Backports should not change on-disk formats or node communication protocols.
• Backports should not add new functionality (except as defined
  here).
• Backports must not add, edit, or otherwise modify cluster versions; or add version gates.
• All backports must be reviewed by the owning areas TL and one additional
  TL. For more information as to how that review should be conducted, please consult the backport
  policy.

If your backport adds new functionality, please ensure that the following additional criteria are satisfied:

• There is a high priority need for the functionality that cannot wait until the next release and is difficult to address in another way.
• The new functionality is additive-only and only runs for clusters which have specifically “opted in” to it (e.g. by a cluster setting).
• New code is protected by a conditional check that is trivial to verify and ensures that it only runs for opt-in clusters. State changes must be further protected such that nodes running old binaries will not be negatively impacted by the new state (with a mixed version test added).
• The PM and TL on the team that owns the changed code have signed off that the change obeys the above rules.
• Your backport must be accompanied by a post to the appropriate Slack
  channel (#db-backports-point-releases or #db-backports-XX-X-release) for awareness and discussion.

Also, please add a brief release justification to the body of your PR to justify this
backport.

[blathers-crl]

Your pull request contains more than 1000 changes. It is strongly encouraged to split big PRs into smaller chunks.

_{🦉 Hoot! I am a Blathers, a bot for CockroachDB. My owner is dev-inf.}

[cockroach-teamcity]

This change is 

[michae2]

Reviewed 8 of 8 files at r1, 2 of 2 files at r2, 14 of 14 files at r3, 10 of 10 files at r4, 1 of 1 files at r5, all commit messages.
Reviewable status: complete! 1 of 0 LGTMs obtained (waiting on @DrewKimball)

[DrewKimball]

Reviewed 8 of 8 files at r1, 2 of 2 files at r2, 14 of 14 files at r3, 10 of 10 files at r4, 1 of 1 files at r5, all commit messages.
Reviewable status: complete! 1 of 0 LGTMs obtained (waiting on @mgartner)

[rytaft]

just a few nits -- feel free to ignore or open another PR on master. This can be merged as-is.

Reviewed 8 of 8 files at r1, 2 of 2 files at r2, 14 of 14 files at r3, 10 of 10 files at r4, 1 of 1 files at r5, all commit messages.
Reviewable status: complete! 2 of 0 LGTMs obtained (waiting on @mgartner)

---

pkg/sql/opt/invertedidx/inverted_index_expr.go line 272 at r3 (raw file):

		con = &constraint.Constraint{}
		con.Init(&keyCtx, &spans)
		break

What if there are multiple similarity filters? Is it worth generating a different scan expression in the memo for each one?

---

pkg/sql/opt/xform/select_funcs.go line 997 at r3 (raw file):

		newScanPrivate := *scanPrivate
		newScanPrivate.Distribution.Regions = nil

nit: maybe add a TODO to calculate the regions here? There are some cases where we should be able to determine the regions statically (e.g., REGIONAL BY TABLE).

---

pkg/sql/opt/xform/testdata/rules/generate_trigram_similarity_inverted_index_scans line 1 at r3 (raw file):

exec-ddl

nice tests!

nit: would help to add some comments about what functionality each test is exercising

[mgartner]

pkg/sql/opt/invertedidx/inverted_index_expr.go line 272 at r3 (raw file):

Previously, rytaft (Rebecca Taft) wrote…

What if there are multiple similarity filters? Is it worth generating a different scan expression in the memo for each one?

Yes, this is naively for now. I've added a TODO here: #122775

[mgartner]

Reviewable status: complete! 2 of 0 LGTMs obtained (waiting on @rytaft)

---

pkg/sql/opt/xform/select_funcs.go line 997 at r3 (raw file):

Previously, rytaft (Rebecca Taft) wrote…

nit: maybe add a TODO to calculate the regions here? There are some cases where we should be able to determine the regions statically (e.g., REGIONAL BY TABLE).

It looks like custom functions never set the distribution directly, and it's done here instead:

cockroach/pkg/sql/opt/xform/optimizer.go

Line 818 in 9514555

provided.Distribution = distribution.BuildProvided(o.ctx, o.evalCtx, relParent, &parentProps.Distribution)

---

pkg/sql/opt/xform/testdata/rules/generate_trigram_similarity_inverted_index_scans line 1 at r3 (raw file):

Previously, rytaft (Rebecca Taft) wrote…

nice tests!

nit: would help to add some comments about what functionality each test is exercising

Done here: #122775