feat(store): iaf migrate-store + dual-store contract suite (epic #540 phase 3d)

[MDUYN]

Phase 3d — iaf migrate-store + dual-store contract suite

Stacked on #546 (Phase 3c). Closes the open Phase 3 deliverables of epic #540.

What's in the box

1. iaf migrate-store CLI

iaf migrate-store --from local-dir   --src ./bt-old \
                  --to   local-tiered --dst ./bt-new

Delegates to dst.copy_from(src), which means the migration is:

• Incremental — bundle by bundle.
• Restartable — re-running over an existing destination just continues.
• Tier-aware — when copying into a local-tiered store, identical OHLCV chunks are written exactly once across the destination regardless of how many bundles reference them (the Phase 3c invariant).
• Selectable — --handles a,b,c for partial migrations.

A migrate_store(...) programmatic helper is also exposed for in-process pipelines.

2. BacktestStoreContractTest — parameterised dual-store suite

A single unittest.TestCase runs an identical scenario against every concrete BacktestStore implementation (LocalDirStore, LocalTieredStore) using subTest(store=label). Adding a future store (e.g. a remote S3BacktestStore) is one entry in _STORES = [...] away from full conformance coverage.

Coverage:

• Protocol + SupportsCopyFrom runtime conformance.
• write → open round-trip preserves algorithm_id.
• summary_only honoured.
• exists reflects writes; idempotent delete removes the handle.
• Missing handle raises StoreHandleNotFoundError.
• iter_handles and __len__ agree with the written set.
• iter_index_rows yields one row per bundle.
• copy_from works for both full migration and a handle subset.

3. Bug fix: LazyOhlcvDict.items() / .values()

The previous implementation inherited the empty backing dict's iteration, so any code path doing

for k, v in bt.ohlcv.items():
    ...

silently dropped every blob after a tiered round-trip — including the migration code we're adding here. Caught by test_ohlcv_dedup_during_migration and fixed by walking the manifest with lazy materialisation, matching the existing __iter__ / __getitem__ semantics.

What is not in this PR

Byte-identical Tier-2 → Backtest reassembly (so .iafbt could become export-only) is intentionally deferred. The current model where the bundle is canonical and Tier-1/2/3 are derived is simpler, preserves the existing round-trip contract bit-for-bit, and is what every test in the contract suite already exercises against both stores. A future slice can promote Tier-2 to authoritative once we have a stronger reassembly story for the envelope's metadata fields.

Tests

• tests/cli/test_migrate_store_command.py — 7 tests (programmatic API + CLI: round-trip, handle subset, OHLCV dedup across migration, unknown-kind validation).
• tests/services/backtest_store/test_store_contract.py — 8 contract tests × 2 stores = 16 subTests.
• Targeted suite (tests/services/backtest_store/ + tests/services/backtest_index/ + tests/cli/): 128 / 128 passing + 26 subTests.
• Full non-scenario suite (tests/ minus tests/scenarios): 1705 / 1705 passing with no regressions from the LazyOhlcvDict fix.

Stack

dev
 └─ #537  feature/bundle-format-v2
     └─ #543  feature/iaf-index-cli
         └─ #544  feature/iaf-backtest-store           (3a)
             └─ #545  feature/iaf-local-tiered-store    (3b)
                 └─ #546  feature/iaf-ohlcv-chunk-store (3c)
                     └─ #???  feature/iaf-migrate-store (3d) ★

Phase 3 of epic #540 is now feature-complete.