[android] - harden offline region creation / update LatLngBounds

[tobrun]

Closes #8514
Closes #4321

[mention-bot]

@tobrun, thanks for your PR! By analyzing this pull request, we identified @zugaldia to be potential reviewers.

[zugaldia]

@tobrun Why are we requiring @UiThread for methods here? I'm thinking of issues if the developer wants to download regions from a background thread, say from a Service.

[tobrun]

I thought this was a requirement as the manager itself must be loaded from main thread afaik, let me fix this up.

[zugaldia]

@tobrun Is it currently possible to get the manager instance from the main thread but then call the other methods in a separate thread?

[zugaldia]

@tobrun Why is necessary to remove this test?

[tobrun]

A sanity test becomes obsolete when there other tests that indirectly test the same thing (this is a common test driven development practice).

[zugaldia]

I always keep those, well, to make sure I keep my sanity ;)

[tobrun]

I'm all down for keeping you sane! Will keep them.

[zugaldia]

@tobrun For all these new methods, could we add specific tests for cases where latitude and/or longitude are crossing their limits (i.e International Date Line)?

[tobrun]

Not sure if there is a reason for it as it's valid to construct a bounds that is not found inside of the world bounds (eg. no wrapping should be enforced by LatLngBounds). The current LatLngBounds logic now matches what core exposes. FWIW this is also the reason why this PR only validates creation of an offline region instead of enforcing this with LatLngBounds.

[zugaldia]

I'm worried about developers using this class outside offline cases. If wrapping is not tested at the core level either, then it's fine to leave it there but I'd make a note in Javadoc.

[tobrun]

added this in class javadoc