Gap: console.time/timeLog timer resolution — prints near-zero instead of real elapsed ms #155

Summary

test_gap_console_methods.ts exercises console.time / console.timeLog / console.timeEnd. The elapsed value prints as 0.005ms / 0.010ms where Node reports 1.271ms / 2.517ms. The actual work between calls is a real setTimeout(resolve, N) so the correct elapsed should be at least the sleep duration.

Actual vs Expected (excerpt)

< timer1: 0.005ms            (perry)
> timer1: 1.271ms            (node)
< timer1: 0.010ms            (perry)
> timer1: 2.517ms            (node)

Scope

• console.time(label) stores the start time at performance.now() precision
• console.timeLog(label) / console.timeEnd(label) compute now - start and format to ms with 3 decimals
• Perry's current impl appears to use a clock that either doesn't advance during the timer wait or uses the wrong monotonic source

Related: see #74 (Date.now not advancing during tight CPU loops — different root cause but similar symptom class).

Impact

One of six currently-failing gap tests (baseline 22/28). Likely 3-4 lines of runtime change in console.rs or builtins.rs.