feat(eap-items): Point eap_items_dlq_replay at eap_items_2 tables (#8575)

Repoints the existing `eap_items_dlq_replay` storage from
`eap_items_1_local` / `eap_items_1_dist` to `eap_items_2_local` /
`eap_items_2_dist`. The main `eap-items-consumer` in us, de and s4s2 now
writes `eap_items_2`, so DLQ replays should land there too.

The schema and `UUIDColumnProcessor` columns are synced with
`eap_items_2`, which adds `conversation_id`, `session_id` and
`ai_conversation_id`. They had drifted from the primary storage. The
storage name, the input topic (`snuba-dead-letter-items`) and the lack
of a `dlq_topic` don't change. Because the name doesn't change,
`eap_items_dlq_grace_period_min` still has no entry for it, so replayed
rows aren't DLQ'd again.

This applies everywhere the replay consumer runs with the default
storage. Any replay consumer that's running when this deploys starts
writing `eap_items_2` on its next restart. No ops change is needed for
regions that use the default storage. This replaces #8570, which added a
separate `eap_items_2_dlq_replay` storage instead.

<!-- junior-request-attribution:start -->
via **Ryan Broughan**.
<!-- junior-request-attribution:end -->

<!-- junior-session-footer:start -->
<!-- junior-conversation-id:slack%3AC0CAC22GMKM%3A1792468188.049119 -->

--

[View Junior
Session](https://junior-prod.sentry.dev/conversations/slack%3AC0CAC22GMKM%3A1792468188.049119)
[[Sentry]](https://sentry.sentry.io/explore/conversations/slack%3AC0CAC22GMKM%3A1792468188.049119/?project=4510944073809921)

<!-- junior-session-footer:end -->

Co-authored-by: sentry-junior[bot] <264270552+sentry-junior[bot]@users.noreply.github.com>
Co-authored-by: Ryan Broughan <ryan.broughan@sentry.io>

Author: sentry-junior[bot]

snuba/datasets/configuration/events_analytics_platform/storages/eap_items_dlq_replay.yaml

@@ -16,6 +16,9 @@ schema:
       { name: item_type, type: UInt, args: { size: 8 } },
       { name: timestamp, type: DateTime },
       { name: trace_id, type: UUID },
+      { name: conversation_id, type: UUID },
+      { name: session_id, type: UUID },
+      { name: ai_conversation_id, type: String },
       { name: item_id, type: UInt, args: { size: 128 } },
       { name: sampling_weight, type: UInt, args: { size: 64 } },
       { name: sampling_factor, type: Float, args: { size: 64 } },
@@ -112,15 +115,15 @@ schema:
       { name: attributes_array_float, type: Map, args: { key: { type: String }, value: { type: Array, args: { inner_type: { type: Float, args: { size: 64 } } } } } },
       { name: attributes_array_bool, type: Map, args: { key: { type: String }, value: { type: Array, args: { inner_type: { type: Bool } } } } },
     ]
-  local_table_name: eap_items_1_local
-  dist_table_name: eap_items_1_dist
+  local_table_name: eap_items_2_local
+  dist_table_name: eap_items_2_dist
   partition_format: [retention_days, date]
 
 query_processors:
   - processor: UniqInSelectAndHavingProcessor
   - processor: UUIDColumnProcessor
     args:
-      columns: [trace_id]
+      columns: [trace_id, conversation_id, session_id]
   - processor: HexIntColumnProcessor
     args:
       columns: [item_id]
@@ -144,7 +147,7 @@ mandatory_condition_checkers:
       field_name: organization_id
 
 # Consumes from the eap_items DLQ topic and writes to the same ClickHouse
-# tables as the primary `eap_items` storage. Intentionally has no
+# tables as the primary `eap_items_2` storage. Intentionally has no
 # `dlq_topic`: messages that fail processing here will fail the consumer
 # loudly rather than getting silently re-DLQ'd back into the same topic.
 stream_loader: