Add failing test: batch scan corrupted in cache_statement mode

[dbussink]

sendBatchExtendedWithDescription passes eqb.ResultFormats to SendQueryStatement by reference. The next Build call reuses the backing array, corrupting the stored format codes for earlier queries. The server sends binary data (Bind was encoded correctly), but the local scan plan uses the corrupted text format code.

Exposed in v5.9.0 by c3a1750 (pgx batch uses SendQueryStatement), part of Skip Describe Portal for cached prepared statements reducing network round trips. The previous SendQueryPrepared path sent Describe Portal, so the server's RowDescription provided correct format codes regardless of the stored slice contents.

@jackc This happened when we updated to 5.9.0. As the test describes, I think it classifies as a bug? I haven't dug into how to fix it yet though, maybe that's immediately obvious to you?

Test fails reproducing the problem we have:

PGX_TEST_DATABASE="host=localhost port=5432 user=postgres sslmode=disable" go test -run TestSendBatchPreservesResultFormats
--- FAIL: TestSendBatchPreservesResultFormats (0.80s)
    --- FAIL: TestSendBatchPreservesResultFormats/cache_statement (0.51s)
        batch_test.go:1290:
                Error Trace:    /Users/dirkjan/code/jackc/pgx/batch_test.go:1290
                                                        /Users/dirkjan/code/jackc/pgx/pgxtest/pgxtest.go:78
                                                        /Users/dirkjan/code/jackc/pgx/pgxtest/pgxtest.go:99
                Error:          Received unexpected error:
                                can't scan into dest[0] (col: int4): strconv.ParseInt: parsing "\x00\x00\x00\x01": invalid syntax
                Test:           TestSendBatchPreservesResultFormats/cache_statement
FAIL
exit status 1
FAIL    github.com/jackc/pgx/v5 1.120s

It shows that it tries to parse the binary wire encoding for 1 as text since the second query in the pipeline uses text.

[jackc]

Good catch. Should have a fix in a few minutes.

[jackc]

Thanks for the great bug report. It was easy to fix with the failing test. Fix released in v5.9.1.

[dbussink]

@jackc Thanks for the quick fix!