locking/atomic: Make test_and_*_bit() ordered on failure

Merge Request Required Information

Summary of Changes

The bug fixed by this commit is being seen on Google Cloud Engine, Ampere and Apple M1 ARM processors.

According to Google, the reproducer attached to the JIRA issue triggers within 60 seconds on affected machines.

Approved Development Ticket

All submissions to CentOS Stream must reference an approved ticket in Red Hat Jira. Please follow the CentOS Stream contribution documentation for how to file this ticket and have it approved.

JIRA: https://issues.redhat.com/browse/RHEL-45199
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>