Clarify Robot Window HTML Test

[CoolSpy3]

Description
This PR fixes a potential race condition in the robot_window_html test. The linked issue mentioned that this test was failing. However, more interestingly, the recent-runs of this test have been passing (since ~Dec 17). No changes were made to the develop branch during this time.

My guess is that this failure was caused by the test not exiting the wb_robot_wwi_receive_text loop after calling wb_robot_wwi_send_text. If the robot window returns the response quickly, it is consumed by the first loop, causing the second loop to hang indefinitely. This would explain the intermittent failures.

This PR forces the loop to break after the send statement, fixing this case. This should hopefully also fix the test on 22.04.

It should be noted that, for me, the test passes normally, but fails in a headless environment (presumably due to the web browser not starting), so that could be contributing to the difficulties in the CI environment. My hope is that because the test has passed before, it should be functional here.

UPDATE: My theory appears to have been wrong. The failure still occurs here (although why it passes on develop remains a mystery. I managed to fix my local build in headless mode by switching off of the default snap install of Firefox, however, that same fix doesn't appear to work here. (In fact, the logs indicate that Firefox is not installed through snap on GitHub runners.)~~

UPDATE 2: Alright, it looks like a runner update was pushed at around the same time this test started passing on develop, so I'm going to guess that that's the culprit. However, it also appears that these changes still fail on master, but pass here. Perhaps it's due to the ongoing work to make develop compatible with 24.04? In any case, I've backed out most of my changes here. What remains is the following:

1. The way the original test was written was somewhat confusing for me. I've rewritten some of the logic to hopefully make it clearer.
   • The test message also referenced complete_test, which briefly led me down a rabbit hole of investigating the ROS tests. I've fixed the message to reflect the correct information.
2. Early on, I started running this test on Ubuntu 22.04. It's been consistently passing there, so I'll assume that whatever problem had led to it being disabled no longer exists. Thus, this PR reenables it.

UPDATE 3: Alright, now the test is consistently failing everywhere. I think it's due to xdg-open not being able to launch a browser in a headless environment. I was able to get around it by explicitly calling firefox -headless, but making that work here would require explicitly invoking Firefox in WbDesktopServices::openUrl, which seems like it would create more problems than it solves. Thus, I think the best course of action is to just disable the test in CI.

Related Issues
This pull-request fixes issue #6727.

[omichel]

Thank you.