sql: flake with distsql_automatic_stats

[yuzefovich]

Seen on this CI run on staging.

Failed
=== RUN   TestTenantLogic_distsql_automatic_stats
    test_log_scope.go:161: test logs captured to: /artifacts/tmp/_tmp/8808ff176222b1d4cd31f20d03de82b0/logTestTenantLogic_distsql_automatic_stats1986543448
    test_log_scope.go:79: use -show-logs to present logs inline
[01:21:48] setting distsql_workmem='27352B';
[01:21:48] setting distsql_workmem='27352B';
    logic.go:2829: 
         
        /home/roach/.cache/bazel/_bazel_roach/c5a4e7d36696d9cd970af2045211a7df/sandbox/processwrapper-sandbox/4242/execroot/com_github_cockroachdb_cockroach/bazel-out/k8-fastbuild/bin/pkg/ccl/logictestccl/tests/3node-tenant/3node-tenant_test_/3node-tenant_test.runfiles/com_github_cockroachdb_cockroach/pkg/sql/logictest/testdata/logic_test/distsql_automatic_stats:25: SELECT DISTINCT ON (column_names) statistics_name, column_names, row_count, distinct_count, null_count
        FROM [SHOW STATISTICS FOR TABLE data] ORDER BY column_names, created DESC
        expected:
            statistics_name  column_names  row_count  distinct_count  null_count
            __auto__         {a}           1000       10              0
            __auto__         {a,b}         1000       100             0
            __auto__         {a,b,c}       1000       1000            0
            __auto__         {b}           1000       10              0
            __auto__         {c}           1000       10              0
            __auto__         {d}           1000       1               1000
            
        but found (query options: "colnames,retry") :
            statistics_name  column_names  row_count  distinct_count  null_count

Jira issue: CRDB-26052

[yuzefovich]

I just hit it quickly when trying to repro #99753. I used ./dev testlogic base --files=distsql_automatic_stats --config=fakedist-disk -v --stream-output --stress --cpus=48 -- --test_arg -test.timeout=10m &> stress.out on the gceworker with 24 cores.

In this case the machine is overloaded, so retry directive doesn't succeed within 45s (the auto stats aren't kicked off, or the stats cache isn't populated). Hm, we do override stats.DefaultRefreshInterval to 1ms in logic tests, so I'm not sure how else we can nudge the refresher. Perhaps the answer is that we should use longer duration for "succeeds soon" retry in the logic tests.