Epic: automation-rate transparency — publish the manual-review rate and gate its reduction

[JSONbored]

Problem

The share of PRs decided fully automatically (no human touch) is a core product metric — the stated target is 97%+ — but it is not computed, persisted, or published anywhere. Progress toward it is currently asserted, not measured.

Goal

A weekly automation-rate series derived from the decision ledger, published on /fairness, with reductions in the manual-review surface gated behind the existing backtest machinery rather than judgment calls.

Sub-issues

• compute + persist the weekly automation-rate series from the decision ledger — shipped via metrics(ledger): weekly automation-rate series, published on /fairness #9938 (closes metrics(ledger): weekly automation-rate series — share of PRs decided with no human touch #9727, ui(fairness): publish the automation-rate series with methodology note #9728)
• publish the series on /fairness with a methodology note — series published via metrics(ledger): weekly automation-rate series, published on /fairness #9938; methodology note added via docs(fairness): write down what the automation rate actually counts #10066
• next backtest-gated guardrail-loosening tranche — this is orb(gate): next guardrail-loosening tranche — backtest-cleared paths only #9729, explicitly still open (docs(fairness): write down what the automation rate actually counts #10066: "This does not close the Epic: automation-rate transparency — publish the manual-review rate and gate its reduction #9726 epic: its third sub-issue is orb(gate): next guardrail-loosening tranche — backtest-cleared paths only #9729, which is blocked on orb(ledger): a hold records reason_code 'success', so the manual-review surface cannot be split by cause #9991" — orb(ledger): a hold records reason_code 'success', so the manual-review surface cannot be split by cause #9991 has since landed via fix(ledger): record WHY a pull request was held, not the gate's conclusion #10072, unblocking orb(gate): next guardrail-loosening tranche — backtest-cleared paths only #9729, but the loosening tranche itself has not shipped yet)

Acceptance

Anyone can see the current automation rate, its trend, and how it is defined — and every future loosening of the manual-review surface has a backtest record behind it.

[linear-code]

JSO-6785