feat(runner): evaluate exit-code install-gate policy into artifact verdict

[jesse-merhi]

New behavior

Config-defined scanners can turn their process exit code into a recorded
pass, warn, or block gate verdict. ClawScan writes the verdict to the run
artifact and prints it in the human scan summary.

• blockOnExitCode and warnOnExitCode accept one code, a list, or nonzero.
• Explicit codes are limited to 0–124. Shell, container-runtime, signal, and
  timeout statuses are not treated as scanner verdicts.
• block wins over warn, and no fired rule records pass.
• Repeated scanner selections execute once, so a later duplicate cannot erase
  an earlier result.
• A CLI --scanner override drops gate policies for excluded profile scanners.
• Invalid, null, empty, aliased-null, or overlapping gate policies fail during
  config resolution.

Gate actions are record-only: block does not stop later scanners or the judge,
and it does not change ClawScan's process exit status. Enforcement belongs to
the caller that consumes the artifact or human summary.

Observable contract

Scanner outcome	Recorded result
Exit code matches blockOnExitCode	gate: block plus one gateRules entry
Exit code matches warnOnExitCode	gate: warn plus one gateRules entry
No rule matches	gate: pass, gateRules: []
Timeout, signal, or exit 125+ with valid JSON	Raw evidence is preserved; exitCode is omitted and no rule fires

For a single run, inspect gate and gateRules. Batch artifacts expose
runs[].gate and runs[].gateRules; benchmark artifacts expose
cases[].run.gate and cases[].run.gateRules. The human scan summary
aggregates the strongest batch action.

How to verify

Create a config-defined scanner that emits JSON and exits 3:

version: 1
profiles:
  demo:
    scanners:
      - id: demo-scanner
        command: |
          test -d {{target}}
          printf '{"findings":1}\n'
          exit 3
        gate:
          blockOnExitCode: 3

Run it without Docker or credentials:

mkdir -p /tmp/clawscan-gate-demo
printf '# demo\n' > /tmp/clawscan-gate-demo/SKILL.md
go run ./cmd/clawscan /tmp/clawscan-gate-demo \
  --config ./with-gate.yml \
  --profile demo \
  --sandbox off \
  --json

The artifact contains:

{
  "gate": "block",
  "gateRules": [
    {
      "scanner": "demo-scanner",
      "rule": "blockOnExitCode",
      "exitCode": 3,
      "action": "block"
    }
  ]
}

Running without --json prints:

gate: block (demo-scanner exit 3 -> block)

The command still exits 0 because this PR records but does not enforce the
verdict.

Checks

• Focused profile, runner, benchmark, and CLI gate regressions passed.
• go test -count=1 ./internal/runner -skip 'TestResolveTargetClassifiesPlugin(Directory|ManifestFile)$' passed.
• All other Go packages passed.
• go vet ./... passed.
• go build ./... passed.
• make docs-site passed in an isolated validation copy.
• The unfiltered suite reaches two unchanged macOS assertions that compare
  equivalent /var/... and /private/var/... paths; the affected production
  and test files are unchanged from main.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[clawsweeper]

ClawSweeper status: review started.

I am starting a fresh review of this pull request: feat(runner): evaluate exit-code install-gate policy into artifact verdict This is item 1/1 in the current shard. Shard 0/1.

This placeholder means the worker is alive and reading the current context. I will edit this same comment with the actual review when the claws are done clicking.

Crustacean status: shell secured, claws on keyboard, evidence pebbles being sorted.