stmtbundle: add recent schema changes and descriptor JSON

[michae2]

When investigating optimizer bugs, In order to rule out schema changes and other interactions with descriptor state, we should consider adding two things to statement diagnostics bundles:

1. a list of all known schema changes for each schema object used by the statement
2. the pretty-printed descriptor JSON for the descriptor of each schema object used by the statement

Jira issue: CRDB-64033

[michae2]

Here's how I usually look at descriptor state:

CREATE TABLE abc (a INT PRIMARY KEY, b INT, c INT, INDEX (b));

SELECT jsonb_pretty(crdb_internal.pb_to_json('cockroach.sql.sqlbase.Descriptor', descriptor, false)) AS d
FROM system.descriptor
WHERE id = 'abc'::regclass::oid;

[virajchogle]

Hi @michae2! I’d like to take this on as my first open-source contribution to CRDB.

Plan: add a new addDescriptors method on stmtBundleBuilder in pkg/sql/explain_bundle.go that walks the opt.Metadata objects addEnv already collects (tables, sequences, views, UDTs, routines) and adds two files to the bundle:

• descriptors.json : one entry per descriptor ID via crdb_internal.pb_to_json('cockroach.sql.sqlbase.Descriptor', descriptor, false, b.flags.RedactValues) over system.descriptor.

• schema_changes.txt : rows from crdb_internal.system_jobs for SCHEMA CHANGE, NEW SCHEMA CHANGE, and TYPE SCHEMA CHANGE jobs, filtered in SQL by decoding the payload and checking for descriptor-ID overlap.

Let me know if there’s anything you’d want done differently before I open the PR.