C3: SessionContext construction in Crawler::process_job

[filipeforattini]

Parent

#46

What to build

Crawler::process_job constructs a concrete SessionContext per Job from the existing Crawler state — proxy router lease, antibot session state, render budgets, resolved policy profile, impersonate profile identity. This is the precursor to the runner cutover: the SessionContext exists at the right point in process_job but is not yet consumed by runner.run.

After this slice, Crawler::process_job builds let ctx = SessionContext { ... }; from self.proxy_router, self.session_states[session_id], self.render_budgets, self.config.policy_profile, and the ImpersonateClient identity. The context is held in a local variable, ready for C4 to thread into runner.run(&job, &ctx).await.

Acceptance criteria

• Crawler::process_job constructs a fully-populated SessionContext per Job
• SessionContext.identity carries the active profile name + locale + session id
• SessionContext.proxy is Some(ProxyLease) when a lease is acquired, None otherwise
• SessionContext.session_state mirrors self.session_states[session_id]
• SessionContext.budgets reflects self.render_budgets remaining values + self.config.max_retries - job.attempts
• SessionContext.policy reflects the resolved PolicyProfile from config
• The SessionContext is constructed but not yet consumed by runner.run (consumption lands in C4)
• No behavior change observable to integration tests; NDJSON regression byte-stable
• All 44 runner unit tests pass; 4 integration tests pass

Blocked by

• C1: JobOutcome typed widening (Vec<Detection> + Option<SessionState>) #47