Previously these would silently fail unless the caller checked the
.error property of the return value. There are no tests that check
iteractions with non-existent methods so this should always be an
error at the test runner level, rather than relying on clients to
check the error.

1. Fix the tests that accidentally call methods that don't exist.
2. Change the test runner so that it prints an error and ends the test.
3. Add a test that the test runner does #2.

Bug: v8:10134
Change-Id: Idd619950a057290c565d58fba6db3ddbcaf2c5eb
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2006093
Commit-Queue: Peter Marshall <petermarshall@chromium.org>
Reviewed-by: Simon Zünd <szuend@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65835}