*: move RowFetcher and co to a new package, row

[jordanlewis]

This was purely mechanical renaming. The purpose is to lessen the width
of the sqlbase package, which is imported in many places.

The new package now depends on sqlbase, but far fewer packages consume
it than sqlbase.

Major changes:

RowFetcher -> row.Fetcher
RowUpdater -> row.Updater
RowDeleter -> row.Deleter

FK and cascade stuff -> row package

The only unfortunate thing here is that the FK enums all begin with
row now (e.g. row.CheckDeletes). It would be better if those got a
different package name, like fk, but I'll leave that refactor for
another time.

Somewhat related to #30001.

cc @dt, @benesch

Release note: None

[cockroach-teamcity]

This change is 

[cockroach-teamcity]

This change is 

[knz]

woo yay to modularity!

minor nits below

Reviewed 50 of 50 files at r1.
Reviewable status: complete! 0 of 0 LGTMs obtained (and 1 stale)

---

pkg/sql/distsqlrun/tablereader.go, line 20 at r1 (raw file):

	"context"

	"github.com/cockroachdb/cockroach/pkg/sql/row"

put the crdb imports below

---

pkg/sql/logictest/logic.go, line 41 at r1 (raw file):

	"unicode/utf8"

	"github.com/cockroachdb/cockroach/pkg/sql/row"

pull below

---

pkg/sql/row/cascader.go, line 30 at r1 (raw file):

	"github.com/cockroachdb/cockroach/pkg/util/log"
)

you could have said type ID = sqlbase.ID somewhere in this package and simplified the diff.

---

pkg/sql/row/fetcher.go, line 23 at r1 (raw file):

	"strings"

	"github.com/cockroachdb/cockroach/pkg/sql/sqlbase"

pull below

---

pkg/sql/row/fetcher_mvcc_test.go, line 23 at r1 (raw file):

	"testing"

	"github.com/cockroachdb/cockroach/pkg/sql/row"

ditto

---

pkg/sql/row/fetcher_test.go, line 23 at r1 (raw file):

	"testing"

	"github.com/cockroachdb/cockroach/pkg/sql/sqlbase"

ditto

---

pkg/sql/row/fk_test.go, line 25 at r1 (raw file):

	"testing"

	"github.com/cockroachdb/cockroach/pkg/sql/sqlbase"

ditto

[jordanlewis]

TFTR! I'm going to fix these import nits, but I'd like to say that it's ridiculous that reviewers have to search for them. Why do we care? More importantly, why doesn't crlfmt do it? :)

[jordanlewis]

TFTR! I'm going to fix these import nits, but I'd like to say that it's ridiculous that reviewers have to search for them. Why do we care? More importantly, why doesn't crlfmt do it? :)

[dt]

I think it was @bdarnell a couple years ago who put it "if it is good enough for goimports, it is good enough for us".

EDIT: found #6096 (comment)

[dt]

I think it was @bdarnell a couple years ago who put it "if it is good enough for goimports, it is good enough for us".

EDIT: found #6096 (comment)

[jordanlewis]

TFTRs!

bors r+

[jordanlewis]

TFTRs!

bors r+

[benesch]

Wow, this looks much easier than what I was trying to do. Thanks Jordan!

[benesch]

Wow, this looks much easier than what I was trying to do. Thanks Jordan!

[craig]

Build succeeded

• GitHub CI (Cockroach)

[craig]

Build succeeded

• GitHub CI (Cockroach)

[danhhz]

I missed the window, but anyway. Nice cleanup.

Ditto dt and ben (and radu's thumbs up). I've started opting out when reviewers nit the import block

[danhhz]

I missed the window, but anyway. Nice cleanup.

Ditto dt and ben (and radu's thumbs up). I've started opting out when reviewers nit the import block