Provide gesture and tappable element insets for caption

Caption is covering the app region and will take gesture and tap events
when the user interact with the caption. This change will make sure the
app can receive the caption as a part of the tappable element insets and
gesture insets to avoid the caption overlaps with interactive elements
inside the window.

Bug: 219987804
Bug: 209717743
Test: atest, see the bugs

Change-Id: I6e48f8df6eb8f73a2f62f34109f4d80d09021929