Java/C#: Add overrides to the interpretation of neutral MaD models.

[aschackmull]

Neutrals generally have little impact on the analysis, but having a neutral like Map.containsKey also cover HashMap.containsKey makes a lot of sense. This might remove a tiny bit of source dispatch for low-confidence dispatch to source implementations of such methods, as that's AFAIR the only touchpoint between neutrals and analysis semantics.

[michaelnebel]

Would you also be up to making the same change for C# as a part of this PR to align the semantics of C# and Java? (if you prefer I can make the corresponding PR for C# instead).
It appears that some tests are failing.

[aschackmull]

Would you also be up to making the same change for C# as a part of this PR to align the semantics of C# and Java? (if you prefer I can make the corresponding PR for C# instead).

Done. And I've fixed the Java tests. What about the failing C# test? Accept the test changes? I haven't tried to dig into exactly what that test does.

[michaelnebel]

Would you also be up to making the same change for C# as a part of this PR to align the semantics of C# and Java? (if you prefer I can make the corresponding PR for C# instead).

Done. And I've fixed the Java tests. What about the failing C# test? Accept the test changes? I haven't tried to dig into exactly what that test does.

Yes, you can accept the updated test output: In this case the updated rows are callables targeted by the neutral MaD rows.

[michaelnebel]

Looks good to me - but maybe also run DCA.

[aschackmull]

but maybe also run DCA

That seems a bit like potential overkill - neutrals really shouldn't have much effect on analysis, but I guess there is a small chance of impact. I'll kick off a pair of runs.

[aschackmull]

Dca was fairly uneventful - mostly just telemetry changes, which makes sense.