keymint-in-vm: Rely on libbinder's new connection retry logic

Previously, a transient connection error when establishing the
accessor service would immediately fail in `libbinder`. This would
require higher-level clients, like KeyMint, to implement their own
retry mechanisms to be robust.

As `libbinder` now handles these accessor connection errors by
retrying internally, this cl leverages that new functionality by
switching from `get_interface` to `wait_for_interface`. This allows
KeyMint HAL to delegate the responsibility of retrying to the
underlying library, simplifying the client code and removing the
need for a custom, high-level retry loop.

Bug: 432582317
Test: Launch Trusty on qemu with Security VM

Change-Id: I5f8f724ed15238c2f8bff5d848719710206b1dec