Make Process.wait safe to retry after timeout

[floitsch]

Summary

• retain one subprocess ResourceState across an interrupted wait
• reuse that state when Process.wait is retried and cache the exit result
• safely dispose pending wait state when wait-ignore is used
• best-effort ignore abandoned waiters through finalization while retaining the explicit wait or wait-ignore contract
• cover repeated timeouts followed by SIGKILL, plus a child that ignores SIGTERM

The complementary SDK-side defensive fix is toitlang/toit#3197. The supported Process termination API is the stacked follow-up #99.

Tests

• toit analyze src/pipe.toit tests/pipe2_test.toit
• toit run tests/pipe2_test.toit
• rebuilt SDK also passed tests/pipe2_test.toit