Fix flaky tests

What does this MR do?

1. Fix logrus race conditions with logger

Tests were using logrus.StandardLogger(), a global singleton, which caused data races when tests ran in parallel with t.Parallel(). Multiple goroutines were simultaneously writing to and reading from the same logger's internal buffer.

Sample failure: https://gitlab.com/gitlab-org/gitlab-runner/-/jobs/12373486144

Solution:

• retry_requester_test.go: Add a test-specific logger using test.NewNullLogger()
• gitlab_test.go: Changed logrus.StandardLogger() to logrus.New() to create a new logger instance for each test run

2. Increase timeout in TestDockerCommandOwnershipOverflow

This test takes about 22 seconds on my local system. It might be taking longer to complete in CI, so extend the timeout from 1 to 2 minutes.

Sample failure: https://gitlab.com/gitlab-org/gitlab-runner/-/jobs/12373486161