Task: Map the remaining stress shift families

Every shift family a shipped stresstest.xml uses round trips, so the stress kind is proven end to end.

• The corpus's fourteen families all map and reverse
• Every shipped stresstest.xml round trips through the database

(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.

All sixteen shift families map both ways, so every shipped stress document round trips. The mapper went from one family to sixteen.

A family keeps four fields in columns of their own: shift_type, shifts, shift_tenors and shift_expiries. Everything else travels in extras as name=value pairs joined by ;, and a nested struct joins its own fields by |. ParShifts is three flags rather than a list, so it becomes one row carrying them in extras. A composite key joins its present parts with |, and a leading | marks a key with no currency.

Three defects were found and fixed on the way. A ParConversion convention id shared the separator of the structure holding it, so the id and every convention after the first were destroyed. The composite key decoder guessed a lone part's meaning from the currency vocabulary, so a key that spelled a currency was silently read as the currency and lost. The library flag was written back with a capitalised boolean, so no document round tripped byte for byte. The build then found a fourth: the configuration_id generator called boost::uuids::nil_uuid() without the header declaring it.

The round trip is proved twice, and the second proof is the load-bearing one. Over the corpus: twenty files, forty-six scenarios, 2047 assertions. Over a document that fills every member of every family at once: 147 checks. The corpus alone would not have found the ParConversion collision, because no shipped file carries that element; the maximal document did.

compass build succeeds, and ores.ore.core.tests, ores.analytics.core.tests, ores.analytics.service.tests and ores.shell.analytics.tests all pass against a fleet restarted on a database recreated from the tree. Merged as PR #2488.

One limit is recorded rather than hidden: a separator embedded inside a data value still corrupts, 45 adversarial probes failing. No realistic value in this domain contains one, so no shipped document is affected. Escaping is a task of its own and the data does not ask for it yet.