Model server load values from scenario input #67

Context

With random.seed supplied, Bastion on athena-sim stops at value getAverageServerLoad at tick 0 (after athena#65 merged). Bastion uses Server Load Average in many Wait(Divide(Server Load Average, 100) * k) pacing expressions. Parent: #45.

Problem

Server load is a client-measured quantity. The simulator does not model server load and must not guess it.

Requirements

• Optional scenario field serverLoad with average, peak (and current only if the catalog has a value for it) as numbers, constant for the run. Choose the smallest shape; document in docs/scenario-format.md and docs/scenario.schema.json.
• Server Load, Server Load Average and Server Load Peak evaluate to the supplied numbers. Without the field the run ends missing_input naming serverLoad (no hidden default).
• Record modeled-not-measured in docs/findings.md: the simulator treats load as a scenario constant; real load varies per tick.
• The trace does not need a change record: it is an input, not a draw.
• Run Bastion with seed and serverLoad supplied and record the next stop in the PR body.

Acceptance criteria

• Tests: each value returns the supplied number; missing field gives missing_input; a Wait expression using load advances by the derived time.
• Docs updated: simulator-interface, scenario-format, findings.
• fmt, clippy -D warnings, cargo test, wasm32 check pass.