release-20.2: assorted pgwire and cli/sql backports

[knz]

Release justification: bug fixes in CLI-only functionality.

Backport:

• 1/1 commits from "sql: fix displaying second zone offsets" (sql: fix displaying second zone offsets #55071)
• 1/1 commits from "cli: handle utf in show tables" (cli: handle utf in show tables #56630)
• 1/1 commits from "pgwire: lazily populate second timezone offset" (pgwire: lazily populate second timezone offset #57265)
• 4/4 commits from "sql/pgwire: fix encoding of int4 and bpchar in tuples " (sql/pgwire: fix encoding of int4 and bpchar in tuples  #61013)
• 1/1 commits from "cli/sql: properly format different times and floats" (cli/sql: properly format different times and floats #62310)
• 1/1 commits from "sql/pgwire: fix encoding of decimals with trailing 0s" (sql/pgwire: fix encoding of decimals with trailing 0s #64022)
• 2/2 commits from "pgwire: only show TZ minutes offset if non-zero" (pgwire: only show TZ minutes offset if non-zero #65063)
• 1/1 commits from "cli/sql: properly format common array types" (cli/sql: properly format common array types #66079)

Please see individual PRs for details.

/cc @cockroachdb/release

[cockroach-teamcity]

This change is 

[knz]

I think I'm missing one PR on account of a pgwire/pgtest test case failing (one that pertains to timestamp)

[knz]

any idea what PR that could be?

[otan]

maybe b450c44 ?

[knz]

the failure is from encodings.json

[rafiss]

it appears that the failing test is now:

--- FAIL: TestUtfName (0.40s)
    test_log_scope.go:158: test logs captured to: /go/src/github.com/cockroachdb/cockroach/artifacts/logTestUtfName939491298
    test_log_scope.go:63: use -show-logs to present logs inline
    sql_util_test.go:259: expected output:
          schema_name | table_name | type  | owner | estimated_row_count | locality
        --------------+------------+-------+-------+---------------------+-----------
          public      | żółw       | table | root  |                NULL | NULL
        (1 row)
        
        got:
          schema_name | table_name | type  | owner | estimated_row_count
        --------------+------------+-------+-------+----------------------
          public      | żółw       | table | root  |                   0
        (1 row)

i guess because #57653 is not in 20.2. looks like the test can be fixed to not expect the locality column

[knz]

ok I'll iterate on this tomorrow or Friday

[knz]

Fixed; RFAL

[rafiss]

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

---

pkg/sql/pgwire/types.go, line 71 at r12 (raw file):

// It is ignored (and can be nil) for types which do not need padding.
func (b *writeBuffer) writeTextDatum(
	ctx context.Context, d tree.Datum, conv sessiondata.DataConversionConfig, t *types.T,

just a theory, but it's possible the issue here is that conv.Location is not set correctly for the EncodingTest

this wouldn't affect later branches because they contain ab19e5c

[knz]

wasn't able to make this work.