GET /indexes returns duplicate rows with the wrong schema's index_definition

[hsusul]

Bug

INDEXES_SQL joins pg_indexes only on index name:

JOIN pg_indexes ix ON c.relname = ix.indexname

Index names are unique per schema, not globally. When two schemas have tables with the same primary-key / index name (very common: users_pkey, dup_idx_pkey), the join matches every schema's pg_indexes row. Because ix.indexdef is in the GROUP BY, GET /indexes returns multiple API rows for one OID, and some of them carry the other schema's index_definition.

Reproduce

create schema private;
create table public.dup_idx (id int primary key);
create table private.dup_idx (id int primary key);

Against the current query, filtering to private.dup_idx_pkey yields 2 rows — one with ON private.dup_idx and one with ON public.dup_idx.

Expected

One row per index OID, with index_definition matching that index's schema.

Suggested fix

JOIN pg_indexes ix ON c.relname = ix.indexname AND n.nspname = ix.schemaname

This is independent of #1108 / PR #1109 (which fixes pg_attribute/indkey matching).

Impact

Studio and any client listing indexes in multi-schema databases silently show wrong DDL / duplicate indexes whenever schemas share naming conventions.