Set zoom for source image as GCzoom when the GC is not autoscalable

[arunjose696]

The calculated image zoom for the source image should be the same as the GCzoom(100) when the drawable of GC is not scalable

Steps to reproduce
Run the snippet in #2504
With source and destination at 1:2 the image would be cropped without this change

Fixes: #2504

[github-actions]

Test Results

  118 files  ±0    118 suites  ±0   10m 42s ⏱️ +14s
4 440 tests +1  4 419 ✅ +2  17 💤 ±0  4 ❌  - 1 
  298 runs  ±0    290 ✅ +1   4 💤 ±0  4 ❌  - 1 

For more details on these failures, see this check.

Results for commit d892d79. ± Comparison against base commit cbf7f7a.

♻️ This comment has been updated with latest results.

[fedejeanne]

Test failures are unrelated: #2516

[HeikoKlare]

Would it be possible to add a test for this to ensure that we do not accidentally run into the same situation again? Something like a simplified version of the snippet in #2504 with a very simple image in which you can easily assert based on specific pixel values if the image was properly sized or not.

[arunjose696]

Would it be possible to add a test for this to ensure that we do not accidentally run into the same situation again? Something like a simplified version of the snippet in #2504 with a very simple image in which you can easily assert based on specific pixel values if the image was properly sized or not.

Added test

[HeikoKlare]

The change looks fine and it properly fixes #2504
I have tested with the snippet provided in that issue.

Thank you also for adding a proper regression test. I only have minor comments regarding that test.

[iloveeclipse]

Caused regression in #2557

[arunjose696]

Caused regression in #2557

This regression is in the test added in the same commit

I think the reason here would be e.gc.drawImage()'s draw request to the OS, might not have completed by the time gcCopy.copyArea() was called, in the test. This was a consideration which was thought of, to add readAndDispatch() call but could see there were no failures in the CI for linux when this PR was made so I decided to skip the call.