[qt5]Fix build failure in linux

[JackBoosY]

Now we don't need to use qmake_all to build qt5 and other modules in linux, it seems to break the disabled parameters during the build. See documentation.

Fix the failure of qt5 module build in linux (caused by #5293):

• qt5-base
• qt5-3d
• qt5-activeqt : only support Windows
• qt5-charts
• qt5-connectivity
• qt5-datavis3d
• qt5-declarative
• qt5-gamepad
• qt5-graphicaleffects
• qt5-imageformats
• qt5-location
• qt5-macextras : only support OSX
• qt5-modularscripts
• qt5-mqtt
• qt5-multimedia
• qt5-networkauth
• qt5-purchasing
• qt5-quickcontrols
• qt5-quickcontrols2
• qt5-remoteobjects
• qt5-script
• qt5-scxml
• qt5-sensors
• qt5-serialport
• qt5-speech
• qt5-svg
• qt5-tools
• qt5-virtualkeyboard
• qt5-webchannel
• qt5-websockets
• qt5-webview
• qt5-winextras : only support Windows
• qt5-xmlpatterns

Related: #6962 #5655 #5362 #3859 #3644.

[JackBoosY]

@Rastaban PR-Eager does not seem to work, and ports that have changed have not been rebuilt.
Can you have a look?

[JackBoosY]

/azp run

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[Rastaban]

Forcing a full rebuild under https://dev.azure.com/vcpkg/public/_build/results?buildId=5553. The CI system will cache results so on later runs if there are no changes beyond what was already tested it just pulls the cached results instead of rebuilding the ports. This caching is global so identical changes in separate PRs will also be found. My guess is that it found results from the windows static build PR and did not actually rebuild everything in this one. Forcing a full rebuild gets around that (although some of the failures may be unrelated flaky ports)

[JackBoosY]

@Rastaban Yeah, I guess that is the reason. It might be better to set a cache for each PR.

[JackBoosY]

@Rastaban Qt5 and its components were built successfully. And 4 ports failed to build(no dependency with qt5):

• bond: shell hash changed. Fix PR: [bond]Update bond stack-install.sh hash. #7293.
• ogdf: the ci system cannot download source.
• vxl: file conflict with qt5-location.
• itk: the built-in gdcm used by itk conflicts with port gdcm.

[Rastaban]

@JackBoosY we have discussed making a separate cache for each PR, but it would cause a full rebuild on the first run which would would mean the test would take 19 hours to complete. Also flaky ports (such as bond/ogdf/vxl/itk) would show up in too many unrelated PRs.

This looks good, Sorry it took so long for me to finish looking at it.