Add in Xvfb vs wayland auto-detection

[sxa]

Implementing the solution I described in #5883 (comment) which avoids the need for labels.

Fixes adoptium/infrastructure#3851

[sxa]

Both of these run through correctly running jdk_beans_0 which uses the display using this PR branch and without using sw.os.rhel.10 in the LABEL field:

UBI10 (Wayland)

https://ci.adoptium.net/job/Grinder/14201/console:

17:39:34  + which Xvfb
17:39:34  which: no Xvfb in (/home/jenkins/.local/bin:/home/jenkins/bin:/usr/local/bin:/usr/bin:/usr/local/sbin:/usr/sbin)
[Pipeline] sh
17:39:37  + weston --no-config --socket=wayland-vfb --backend=headless-backend.so --xwayland
17:39:37  Date: 2025-08-28 UTC
17:39:37  [15:39:37.086] weston 14.0.1
17:39:37                 https://wayland.freedesktop.org/
17:39:37                 Bug reports to: https://gitlab.freedesktop.org/wayland/weston/issues/
17:39:37                 Build: 14.0.1
17:39:37  [15:39:37.086] Command line: weston --no-config --socket=wayland-vfb --backend=headless-backend.so --xwayland
17:39:37  [15:39:37.086] OS: Linux, 6.8.0-1018-azure, #21~22.04.1-Ubuntu SMP Fri Nov  8 00:21:25 UTC 2024, x86_64
17:39:37  [15:39:37.086] Flight recorder: enabled
17:39:37  [15:39:37.086] Starting with no config file.
17:39:37  [15:39:37.651] Output repaint window is 7 ms maximum.
17:39:37  [15:39:37.651] Loading module '/usr/lib64/libweston-14/headless-backend.so'
17:39:37  [15:39:37.657] Registered plugin API 'weston_windowed_output_api_headless_v2' of size 16
17:39:37  [15:39:37.657] Color manager: no-op
17:39:37                   protocol support: no
17:39:37  [15:39:37.657] Output 'headless' attempts EOTF mode SDR and colorimetry mode default.
17:39:37  [15:39:37.658] Output 'headless' using color profile: stock sRGB color profile
17:39:37  [15:39:37.658] Output 'headless' enabled with head(s) headless
17:39:37  [15:39:37.658] Compositor capabilities:
17:39:37                 arbitrary surface rotation: no
17:39:37                 screen capture uses y-flip: no
17:39:37                 cursor planes: no
17:39:37                 arbitrary resolutions: no
17:39:37                 view mask clipping: no
17:39:37                 explicit sync: no
17:39:37                 color operations: no
17:39:37                 presentation clock: CLOCK_MONOTONIC_RAW, id 4
17:39:37                 presentation clock resolution: 0.000000001 s
17:39:37  [15:39:37.658] Loading module '/usr/lib64/weston/desktop-shell.so'
17:39:37  [15:39:37.658] Loading module '/usr/lib64/libweston-14/xwayland.so'
17:39:37  [15:39:37.659] Registered plugin API 'weston_xwayland_v3' of size 32
17:39:37  [15:39:37.659] Registered plugin API 'weston_xwayland_surface_v2' of size 24
17:39:37  [15:39:37.659] xserver listening on display :0
17:39:37  [15:39:37.659] launching '/usr/libexec/weston-keyboard'
17:39:37  [15:39:37.660] launching '/usr/libexec/weston-desktop-shell'
[Pipeline] echo
17:39:38  env.DISPLAY is :0

Noting that there was one test which failed in 1/10 iterations in that run: https://ci.adoptium.net/job/Grinder/14202/testReport/java_beans_XMLEncoder_Test4822050/java/Test4822050/ although the stderr output reported STATUS:Passed. The console log showed 20:34:02 java.awt.AWTError: Can't connect to X11 window server using ':0' as the value of the DISPLAY variable. when that occurred (iteration 9/10) but since this is using the original Xvfb startup I believe this is an intermittent glitch and not related to the PR.

UBI9 (X11)

https://ci.adoptium.net/job/Grinder/14202/console:

17:53:50  + which Xvfb
17:53:50  + /usr/bin/which --tty-only --read-alias --read-functions --show-tilde --show-dot Xvfb
17:53:50  + alias
17:53:50  + eval declare -f
17:53:50  ++ declare -f
17:53:50  /usr/bin/Xvfb
[Pipeline] wrap
17:53:50  $ Xvfb -displayfd 2 -screen 0 1024x768x24 -fbdir /home/jenkins/workspace/Grinder/.xvfb-14202-..fbdir17001837949309922997
[Pipeline] {
[Pipeline] sh
17:54:29  + grep '[X]vfb'
17:54:29  + ps -f
17:54:29  + awk '{print $9}'
[Pipeline] echo
17:54:29  env.DISPLAY is :0

RHEL10: https://ci.adoptium.net/job/Grinder/14204/

[sxa]

@smlambert if you're ok with this it would be good to have your approval today since I believe Canada is not around on Monday and we're very close to being able to make UBI10 containers live in jenkins.

[llxia]

I would suggest explicitly checking the return code:
if (sh(script: "which Xvfb 2>&1", returnStatus: true) != 0) {

[sxa]

My preference would be to leave it as-is for now as it's been fully tested and I want to try and get this live, but I'm happy to change it in in a future PR if you're prefer. Let me know if you're ok with that.