[AZ-559] [AZ-389] docs: defer AZ-389 to AZ-559 (C6 mid-flight tile gap)

AZ-389's task spec assumed the existence of `tile_store.put_mid_flight_
candidate(MidFlightTileCandidate)` (in Excluded: "owned by AZ-303 / E-C6"),
but the current TileStore Protocol has only the four-method baseline
shipped under AZ-303 — there is no put_mid_flight_candidate, no
MidFlightTileCandidate DTO, and no MID_FLIGHT_INGEST TileSource enum value.

Filed AZ-559 as a 5pt task to close the C6 storage gap (Protocol method
+ DTO + enum + persistence + freshness/LRU integration + contract
update). Updated AZ-389 spec to depend on AZ-559 (replacing the stale
AZ-303 dep) with a Status: BLOCKED note. Updated the dependencies
table totals: 151 tasks / 502 complexity points.

This is the same dep-gap pattern surfaced for AZ-401 in batch 61
(missing AZ-400 transport-seam retrofit) — the autodev replay-track
sequence is exposing under-spec deliveries upstream. Tracker remains
the source of truth via the new AZ-559 issue + Blocks link.

Co-authored-by: Cursor <cursoragent@cursor.com>

_docs/_autodev_state.md

@@ -15,4 +15,4 @@ tracker: jira
 last_completed_batch: 62
 last_cumulative_review: batches_58-60
 current_batch: 63
-current_batch_tasks: ""
+current_batch_tasks: "AZ-389"