Fix #3946: Fix solution automatically revealing after first hint reveal

[BenHenning]

Explanation

Fix #3946

Per #3946 there's a bug in some contexts where, after revealing a hint, the solution may automatically be revealed as well when the user reopens the solution dialog.

The root cause of this is that the hint handler decides whether to show solutions based on the presence of a correct answer, not based on the presence of a solution itself. This presents a problem specifically for proto-based lessons because in some cases correct_answer will be absent (if the converter can't properly convert the answer into an app-acceptable format, e.g. text or fraction). This is technically a valid case to introduce a failure, but due to #1050 we can't actually assume that correct_answer is properly defined. Thus, the solution is to change the check to instead check for whether solution is present. This does allow for an empty solution object to be added to the proto, but this shouldn't realistically happen in practice so it should be fine.

To repro (on this branch):

• Enable proto loading in CachingModule
• Build & install a dev flavor of the app
• Start a new Prototype Exploration play session
• Play to the numeric expression state
• Enter 2 wrong answers
• Reveal the hint
• Close & reopen the dialog
• (If you want to repro the issue, you'll need to replace the implementation of HintHandlerProdImpl with the current develop copy)
• Without the fix, the solution will automatically be revealed. With the fix, the solution won't yet be available to reveal.

Note that this doesn't affect json since we've taken some measure to ensure that correct_answer is filled with something even if that value isn't quite correct.

To ensure this regression is properly caught, tests were added both in StateFragmentLocalTest and HintHandlerProdImplTest. Note that the latter required introducing a new test exploration that specifically tests a missing correct_answer, and the former required changes to the prototype exploration to introduce a hint & solution in the numeric expression state to more easily repro this issue. Test results:

• Failing without the fix (Bazel only since it requires textproto loading): https://app.buildbuddy.io/invocation/da48df48-444c-48d5-870f-dec2c75b68b1
• Passing with the fixes: https://app.buildbuddy.io/invocation/92f3739b-0477-43c1-82df-8170b8fd197a

Other miscellaneous changes:

• StateFragmentLocalTest had some cleaning up since its helpers were a bit confusing referencing states for different explorations (now that the test suite uses at least 2 explorations for testing)
• StateFragmentLocalTest was updated to depend on textproto assets in Bazel for better coverage, and so that the failure case could be triggered per the new regression tests
• StateRetriever was updated to allow a missing correct_answer in solutions (for parity with the textproto variant, but the implementation is a bit more permissive and actually results in passing tests regardless of the fix)
• StateRetriever was also changed so that solutions are not set if they aren't loaded (to avoid an empty solution being included which shouldn't actually ever happen for loaded binary protos). This ensures that the updated check (hasCorrectAnswer -> hasSolution) results in consistent & correct behavior for both JSON & textproto runs of the hint tests/app builds.
• HintHandlerProdImpl was updated to use an AtomicInteger for hint sequencing. This isn't related to the original issue, but there's a likely race condition in the current implementation that I'm fixing at the same time. No regression test is reasonably possible to actually test this, so it needs to be a verify-on-inspection type change.

Essential Checklist

• The PR title and explanation each start with "Fix #bugnum: " (If this PR fixes part of an issue, prefix the title with "Fix part of #bugnum: ...".)
• Any changes to scripts/assets files have their rationale included in the PR explanation.
• The PR follows the style guide.
• The PR does not contain any unnecessary code changes from Android Studio (reference).
• The PR is made from a branch that's not called "develop" and is up-to-date with "develop".
• The PR is assigned to the appropriate reviewers (reference).

For UI-specific PRs only

N/A -- this is fixing a UI-affecting issue, but doesn't actually change the UI.

[BenHenning]

@rt4914 PTAL (for codeowners + other changes if you can since the PR isn't affecting too many files).

[rt4914]

LGTM, thanks.

[oppiabot]

Hi @BenHenning, this PR is ready to be merged. Please address any remaining comments prior to merging, and feel free to merge this PR once the CI checks pass and you're happy with it. Thanks!

[BenHenning]

Thanks @rt4914!

[BenHenning]

SIGSEGV in StateFragmentTest for the Gradle run. Re-running since that test should pass in this PR.

[BenHenning]

Looks like everything is passing, so merging now.

[BenHenning]

This is confirmed as fixed in the release branch.