[stefan-jansen]

Closes #103

Outcome

Use one internal daily session-label projection across gap classification, range planning and actual backfill provider-row selection. Aware labels restore declared timezone after canonical UTC storage; naive labels remain civil identifiers. Explicit event_time/bar_close/bar-side metadata uses native SessionAssigner even at midnight or daily frequency. Out-of-session instants fail clearly, while legacy daily labels retain their existing off-session policy. Explicit disabled calendars bypass schedule interpretation.

The utility preserves actual bracketing observation timestamps in DataGap. Range bounds remain inclusive session labels at UTC midnight; missing_periods/size_days count expected sessions. Backfill selects original provider rows by projected session label without retimestamping or manufacturing observations. A declared intraday frequency cannot be silently treated as daily observations. No parallel calendar resolver or event-interval implementation.

Executed evidence

• Basef57c8d91fd5dbada151fb4dfe8325bda8268b954, regression-only4c1c0db41a9bb3cdca0787e159dea9092b965820:10 failures/4 controls in1.34s. Real SyntheticProvider labels -> Flat/Hive -> canonical UTC read -> DataManager metadata -> daily gap utility falsely loses Tokyo dates and reports missing Friday. The existing range planner is a passing timezone control. Explicit instant rejections are capability failures, not10 historical numerical bugs.
• Current a78036e includes main 1501f00 through normal expected-head-guarded GitHub Update branch. All six owned files are unchanged from c9b39dd. Before the strict local hold, 6e56b18 passed 176 focused daily/gap/range/update/session cases in 15.83s with ResourceWarning as an error and native thread caps of one. Those historical results do not replace new-head hosted qualification; no local tests or hooks have run during the hold.
• Current-head hosted coverage run 39059906046 passed 3,950 tests, 292 deselected, with 86.24% coverage against the required 85%. Strict documentation run 39059906071 passed its build, rendered identity checks and artifact upload. Remaining supported native/ecosystem platform checks must pass before merge.
• Includes actual Flat/Hive backfill for event/close and Tokyo labels, preserved provider timestamps, continuous midnight sides, actual missing session, disabled calendar, outside-session rejection, custom timestamp column and conflicting frequency. The previous unsupported-event guard is replaced by public session-mapping expectations rather than retained as an obsolete rejection.
• Ruff lint/format, ty, diff checks, normal commit hooks and strict MkDocs pass. Full applicable hosted CI/docs/package/resource/platform gates remain required after prerequisite/retarget; focused evidence alone is not full qualification.

Dependency and review

Targets main with merged PRs 94, 98, 102 and 106 integrated. Retargeting and non-rewriting synchronization are complete; every applicable main-target workflow must qualify this exact combined head. PR 100's separate intraday utility/docs changes still require normal conflict resolution after the hold. They are not folded into this focused review, and old stacked trigger exclusions are not counted as qualification.

AI-assisted source/tests/prose under maintainer direction. Owner reviewed actual diff, shared interpretation, old label policy, timestamp/session units, range bounds, real storage/backfill and immutable inputs. No independent reviewer claimed. No release/deploy or change to published0.2.1. Remaining validation and intraday range-grid requirements remain explicit; native interruption parity is implemented in merged PR94.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.