Task: Parse the engine log when a run fails

A failed run's errors are the parsed structured messages, one per failure, with the trade id and type, not the raw transcript.

• The failure path runs the output through the canonical log reader and records the parsed errors.
• The run's message carries the summary; the Errors tab lists each structured failure.
• A failed run from the demo shows the parsed rows; the tests and gates are green.

(Implementation strategy. Written when work starts; key decisions are distilled into the parent story's * Decisions at close, but the plan itself stays — it is the historical record of what we did.)

Every number this task claims about a tree, with the command that produced it and the commit it was measured at. A number without both is not a measurement, because a clean-up record describes a tree and trunk moves under a branch: a consumer count, a gate's component count and a suite's size all change with no code changing, and the reader of a bare number cannot tell which tree it came from. Where the raw output matters, put it in an evidence/ file beside the task and name that file in the Command column.

Manual QA scenarios (scaffolded via compass add test_scenario) that verify this task. Link new ones here as they're created; the scenario doc itself links back via its "Verifies task" field.

A failed run's errors are now the parsed rows, not a transcript. Two defects were fixed in series: the compute wrapper uploaded its log only on success, so the reporting read a 404; and the failure path recorded the platform's raw message first, which the write-once recorder then protected, so the parsed rows could never land. The wrapper now writes its rows for every job that reached the engine (the structured log first, the text log otherwise), and the failure path records the log before it refuses; the run's message is one sentence that names the count and points at Errors.

Verified live on the rebuilt services: the Headline Position run cd8082af-7f06-4365-a87e-3afc94a7a34d failed and its errors read 40 rows, 39 trade failures plus the run's pointer line, each with its trade id and the engine's message; the run's message reads "The run's compute job failed with outcome 3: Engine exited with code 255. Its log records 39 errors; the rows are in Errors."

ores.reporting.core.tests 267 cases / 1499 assertions pass; compass lint passes; check_component_drift reports no drift for reporting or compute-cpp. A successful run remains unproven: the demo's package lacks Input/simulation.xml, so the engine aborts before it prices, which is the report-inputs task.