[ramondalmau]

Text only; no code or behaviour change.

• Empty AO values (fix(interpolation): skip empty values in operator columns when resampling (audit D4) #81): mrv-specification.md and input-prioritization.md said an empty
  value empties its neighbours and drops out of the fuel totals. It is interpolated in time
  between the nearest points with a value, as input-format.md already says.
• Fuel flow derived from the AO mass (fix(performance): fuel flow derived from aircraft_mass is no longer one point late #79): describe the current formula (mass decrease to the
  next point / time to the next point).
• output-format.md rewritten around runner.results: success entries, low-altitude zero
  results and error records (order, keys, examples that run), fleet_meta_data keys from
  report.py, bada_version / bada_code absent on the AO fuel flow + efficiency + mass path,
  multi-engine entries (DataFrame input with a list engine_uid is not supported: tracked as
  DF-ME). The per-segment field list moves to a "Golden dump" section.
• flight_schema.json, description strings only: hydrogen content wt.%, calorific value
  J/kg, departure/arrival times as supplied, EDB engine UID, load factor basis, whole-aircraft
  fuel flow, propulsion efficiency, flight level, TAS in m/s (AO sends knots).
• usage.md: working ZarrPaths and DWDFactory snippets. examples/README.md and
  example-scripts.md: units and the efficiency definition.
• For Philippe: EAGWP/AGWP units in the MRV formulas corrected from [W m-2] to
  [W m-2 yr]; hydrogen content formula now gives wt.% (x 100).

Checks. Reviewed by independent checkers over 5 rounds against the code; every code example
in output-format.md and the changed usage.md snippets were run. Schema still valid; the 11
golden inputs still validate against it. Non-weather suite: 407 passed.

Still to do (D-b): input-format.md field definitions from the AO schema.

Review. Ramon; Philippe for the two formula changes.