wasm-runner: thread timeout option through all WasmRunner APIs

[yevbar]

By Sleepy

Summary

Previously, WasmRunner functions (run/2, pipe/3, run_batch/3, run_concurrent/4, benchmark/4) had no way to control the per-call execution timeout — the 5000ms default from Runtime.call was always used. This PR threads a :timeout option through the entire WasmRunner API.

Changes

• Firebird.call/4: Now accepts an optional opts keyword list (with :timeout) forwarded to Runtime.call/4
• call_single/4, call_single!/4: Accept optional opts with :timeout
• run/2: Recognizes :timeout in the keyword API (run("m.wasm", add: [1,2], timeout: 10_000))
• pipe/3, run_batch/3, run_concurrent/4: All pass :timeout through to calls
• benchmark/4, benchmark_compare/5: Thread timeout to measured calls
• extract_call_opts/1: Clean separation of call-level opts (:timeout) from start-level opts (:wasi, :cache)
• 12 new tests covering timeout threading for all WasmRunner functions

Usage

# Set 10s timeout for slow WASM functions
{:ok, result} = WasmRunner.run("heavy.wasm", compute: [big_input], timeout: 10_000)

# Timeout in pipe chains
{:ok, result} = WasmRunner.pipe("math.wasm", [{:fib, [40]}], timeout: 30_000)

# Timeout in concurrent execution
{:ok, results} = WasmRunner.run_concurrent("math.wasm", :fib, args, timeout: 15_000)