fix(wasm): route clock reads through web-time so bashkit runs on wasm32-unknown-unknown

[a0preetham]

Problem

std::time::{Instant, SystemTime} panic at runtime on wasm32-unknown-unknown ("time not implemented on this platform"): the target has no OS clock, and std provides no extension point (library/std/src/sys/time/unsupported.rs). Any script touching time — date, touch, ls -l, time, exec duration tracking — aborts when bashkit is embedded in a browser via wasm-bindgen.

Fix

All wall-clock reads in the bashkit crate now go through a new crate::time_compat module:

• on wasm32 it re-exports web-time (Performance.now()/Date.now() via web-sys)
• everywhere else it re-exports std::time — zero overhead, and the dependency is only pulled in for wasm32 targets
• two small bridge helpers cover chrono conversions (chrono implements From<std::time::SystemTime> but not web-time's)

Also gates Interpreter::clear_cancelled_execution_state to non-wasm: its only caller is the native-only tokio timeout recovery path, so it was a dead-code warning on wasm targets.

Guards so this can't rot

• new CI job: cargo check -p bashkit --target wasm32-unknown-unknown --features scripted_tool,jq with -D warnings (python is excluded: monty → ahash → getrandom 0.3 lacks its wasm_js feature there — separable follow-up)
• new source-scan test (same pattern as release_profile_tests.rs) keeping std::time clock types out of src/, with a // std-time-ok: <reason> escape hatch

Verification

• full just check equivalent passes (fmt, clippy -D warnings, workspace tests + doctests)
• cargo check --target wasm32-unknown-unknown passes warning-free
• verified end-to-end in a browser: a wasm-bindgen build of bashkit on this branch runs in headless Chromium with date -u returning the actual current time, and touch/ls -la showing correct timestamps

[chaliy]

This loses subsecond precision for timestamps before UNIX_EPOCH. For example, 1969-12-31T23:59:59.500Z should round-trip as 500ms before epoch, but this path converts using whole negative seconds and zero nanos. Please add focused tests for pre-epoch subsecond values and decompose negative timestamps so nanos are preserved in both to_chrono_utc and from_chrono.

[a0preetham]

Thanks for quick review. Done!