fix: enforce uniqueness and hour alignment for agent runtime usage events - #27983

…ents

The usage generator writes hb_agent_runtime_v1 rows with created_at at
the UTC hourly bucket start and exactly one row per bucket, but nothing
in the schema enforced either invariant. A duplicate bucket row under a
different id would be double-counted by any consumer summing runtime_ms,
and a misaligned created_at would skew which usage period a bucket is
attributed to.

Replace the non-unique partial index idx_usage_events_agent_runtime with
a unique index of the same shape and add an hour-alignment CHECK
constraint. generateBucket treats a unique violation on the bucket index
as another replica having won the race, mirroring the existing
ON CONFLICT (id) no-op for committed rows.

…mplate goldens with generators"

This reverts commit ed531d9d365720b4e394f4b422ea9776886c56a3.

The gen/fmt drift it patched over came from main commit 07f79af
(fix: markdown rendering improvements) having landed out of sync with
its generators; main has since reverted that commit entirely in
8c2f7ad (#27979), so the sync is obsolete and conflicts with the
regenerated goldens on latest main.

The expanded arbiter-scoping commentary is not needed on the insert
query; the duplicate-bucket behavior is documented on the migration and
in generateBucket where it is handled.

…000567

Main landed its own 000565 and 000566 (oauth2 client type constraint and
auth method backfill) after this branch was cut, making the migration
number a duplicate that fails migrate setup.

…e comment

…000569

Main landed its own 000567 and 000568 (chat file purge indexes and
service account notifications) after this branch was renumbered to
000567, making the migration number a duplicate that fails migrate
setup in the test merge.

The (id) arbiter is not limited to committed rows: when a competing
row's arbiter index entry is visible, PostgreSQL waits on that
transaction and takes the DO NOTHING path if it commits. Only the
narrower speculative-insertion race, before the arbiter entry exists,
trips the bucket unique index.

…tp test

The revert in cb36cfa left behind an unused
"github.com/coder/coder/v2/coderd/render" import, breaking compilation
of the dispatch test package and failing fmt, lint, and every
test-go-pg job. Restore the import block to match main.

…000570

Main landed 000569_oauth2_scope_columns after the previous renumber to
000569, so the CI test merge contained two migrations with the same
number and every migration-dependent job failed. dump.sql is unchanged
because only the file number moves.