fix(android): fix views with border and transparency

[ypbnv]

JIRA: https://jira.appcelerator.org/browse/TIMOB-26394

Description:
Rework handling the disabling of the HW acceleration for views with transparency and borders. isHardwareAccelerated() returns whether the view is attached to a Window with enabled HW acceleration:
https://android.googlesource.com/platform/frameworks/base.git/+/refs/heads/master/core/java/android/view/View.java#20915
So it depends on the view being attached to a Window. Our call of disableHWAcceleration() in TiUIView.java happens before the native view is actually attached to a window, so it would require an additional check in order to determine when to disable acceleration for the TiBorderWrapperView. Currently borderView.isHardwareAccelerated() will always return false, resulting in skipping our method.
We had a change in 6.3.0.GA for this code in this PR:
#9410
Ran the test cases from it and it works fine both on Android 19 and Android 29.
This PR also fixes the case described in this JIRA ticket - https://jira.appcelerator.org/browse/TIMOB-26291
During FR I would recommend trying the test cases from all three PRs to avoid regressions.
Note: No unit test, since this is a visual fix.

Test case:
app.js

var win = Ti.UI.createWindow({
     backgroundColor: '#fff'
});
 
var v1 = Titanium.UI.createView({
     backgroundColor: 'rgba(0,0,0,0.5)',
     height: 100,
     width: Ti.UI.FILL,
     borderRadius: 20
});
win.add(v1);
win.open();

[build]

	Messages
📖	💾 Here's the generated SDK zipfile.
📖	✅ All tests are passing Nice one! All 3899 tests are passing. (There are 475 skipped tests not included in that total)
📖	✊ The commits in this PR match our conventions! Feel free to Rebase and Merge this PR when ready.

Generated by 🚫 dangerJS against 30ff261

[garymathews]

CR: PASS

[ssjsamir]

FR Passed: Views with borderRadius now show up semi-transparent. Tested with the test case in the description and additional tests from the mentioned PRs above were done.

Test Environment

MacOS Catalina 10.15 Beta
Node.js ^8.11.1
Android Pixel XL 7.1 Emulator 
"NPM":"4.2.15","CLI":"7.1.1"
Java Version 1.8.0_131
Android NDK:  12.1.2977051

[hansemannn]

Either this one or #10637 is causing a regression with Ti.Map which now displays an empty map, after updating to latest master. You can try with our internal project, I think both @garymathews and @jquick-axway have access to it.

[jquick-axway]

@hansemannn, we'll check it out. Thanks.
Question: Do you have a border applied to your map?

[jquick-axway]

Okay, I've confirmed that we have a regression on "master". If you apply a border to a VideoPlayer or MapView (something that requires hardware acceleration), then it won't render. I can understand this happening if "borderRadius" was involved (that's buggy on Google's end), but this is happening to rectangular borders now.

var window = Ti.UI.createWindow({ fullscreen: true });
var videoView = Ti.Media.createVideoPlayer({
	url: "http://assets.appcelerator.com.s3.amazonaws.com/video/media.m4v",
	autoplay: true,
	borderWidth: "4dp",
	borderColor: "red",
	width: Ti.UI.FILL,
	height: Ti.UI.FILL,
});
window.add(videoView);
window.open();

I've written up a private ticket for it here: TIMOB-27444

[jquick-axway]

The fix for the regression can be found here:
#11262

It turns out the old code (before this PR) was wrongly "trying" to disabling hardware acceleration when it didn't need to, but failing due to what this PR was fixing. So, this PR merely brought the pre-existing bug to the surface.

[hansemannn]

Thanks Josh, that was fast!