fix(humanize): share timeout budget in select_option (follow-up to #307)

[Kumario1]

Summary

Follow-up to #307.

58ccdb6 fixed the multi-wait timeout bug (a 5s timeout waiting ~15s) for
click / dblclick / hover / dragAndDrop on frame and ElementHandle by
sharing a single deadline + remainingMs() budget. The select_option
paths were missed and still have the same class of bug.

Each select_option helper computes a shared deadline for the actionability +
hover steps, but then calls the native select_option with the original
full timeout instead of the remaining budget — so the native call restarts the
clock. A slow <select> can therefore wait up to ~2× the requested timeout.
The frame-level wrapper didn't even compute a deadline (full timeout to both the
hover and the native select).

Changes

Thread the remaining budget into the native call across every variant:

• page _human_select_option (sync + async)
• ElementHandle _human_el_select_option (sync + async) — both the
  move-failed early-return and the normal path
• frame _frame_select_option (sync + async) — add a shared deadline
• Locator _humanized_select_option (sync + async) now delegates to
  page.select_option (like click/hover/fill already do), so there's a
  single source of truth for the budget instead of a second hover + native wait

Testing

python -m pytest tests/test_humanize_unit.py -k TestSelectOptionTimeoutBudget
# 3 passed

New TestSelectOptionTimeoutBudget:

• page native select_option receives the remaining budget (< 5000), not the full timeout
• Locator.select_option delegates to page.select_option with the forwarded timeout
• frame select_option shares one deadline across hover + native select

Each test fails against the pre-fix behaviour. Full tests/test_humanize_unit.py
run: 76 passed; the only failures are 5 async tests that also fail on a clean
main here (the env has no pytest-asyncio configured) — unrelated to this change.