queen: service levels with an error budget -- written targets, and dispatch slows when review falls behind #9974

Part of the dispatch roadmap epic #9976. Industry reference and the measured gap are below.

Measured 2026-10-17 14:37Z: 45 dispatched/h but 24 reviewed/h and 21 finished/h; 206 unreviewed. Dispatching faster than review can absorb only grows the queue and the holds (specs/queen/holds.t27). No target is written for idle lanes, review lag, unexpected exits or lead time, so no rule can act on them.

Industry reference: Google SRE SLOs with error budgets; Kanban WIP limits and Little's law (lead time = WIP / throughput); DORA lead time and change-failure rate.

User Scenarios

• As the owner, I read one card with the swarm's targets: idle lanes while eligible work exists, review lag, unexpected-exit share, lead time p90, change-failure rate.
• As the Queen, when unreviewed work exceeds what review clears in a set number of hours, I lower dispatch until it recovers instead of filling lanes with work that will only wait.

Requirements

• Named targets as constants with the measured baseline in the comment: idle lanes 46% -> < 5% while eligible work exists; unexpected exits up to 78% -> < 2%; unreviewed 206 -> < 30; lead time p90 49.6 h -> < 6 h.
• dispatch_allowance(free_lanes, unreviewed, reviewed_per_hour) -> u32 (a Little's-law WIP cap) and budget_left(events, bad_events, slo_permille) -> i32.

Success Criteria

• Tests at the baseline numbers and at the targets; the allowance never exceeds free lanes and never drops below one when review is healthy.
• 0 surviving mutants.

Rules for the bee

• Only .t27 is hand-written; the seal is tool-written (t27c seal --save after a fresh release build on the Railway lab). No .py, .rs, .ts, .sh or .yml edits.
• Every rule gets test blocks and at least one invariant; run tri mutate spec on the new file (on the lab) and kill every surviving mutant. The bar is specs/queen/dispatch.t27 in PR test(queen): kill the 39 surviving mutants of specs/queen/dispatch.t27 #9961: 0 survivors, 0 vacuous tests.
• Write the measured numbers that motivate the card into its header comment, as specs/queen/holds.t27 does.
• Wiring into the runtime (gHashTag/trios) is a separate step: it lands as generated code once gen-ts lowers pure function bodies (gen-ts: lower the bodies of pure functions, so a host imports a spec's rule instead of copying it #7871 / PR gen-ts: lower the bodies of pure functions #7872). Do not hand-write TypeScript here.

Boundary

• specs/queen/slo.t27
• .trinity/seals/queen_QueenSlo.json