under -source:future, don't warn on infix usage of Scala 2 defined methods with default arguments

[SethTisue]

fixes #10383

this is WIPpy at the moment, but plausible, and we wanted to get CI feedback over the weekend :-)

[dwijnand]

Looks like Apply is a ProxyTree, so it delegates its denotation to the underlying function, while Block isn't, so its denotation is NoDenotation - which doesn't have Scala2x...

So another way, or perhaps in addition, would be to verify the symbol exists before deciding whether the method is Scala 2 or not. Or assert it exists.

[SethTisue]

Aside:

We also tried making Block forward its denot method to expr and that did fix the bug without having to change Typer, and in testCompilation only three failures ensued. And we tried going even a little farther and making Block actually extend ProxyTree and there were also only a handful of failures (nearly the same set).

We looked at those failures for a while out of curiosity, but in the end, changing Block in either of those ways would be a pretty fundamental change, and even if the test failures are surely fixable, it seems like too ambitious a change in the context of this bug.

If any reviewer happens to know why a Block isn't a ProxyTree, we're all ears — the low number of test failures when we make it one seems to indicate that it plausibly could be.

[SethTisue]

So another way, or perhaps in addition, would be to verify the symbol exists before deciding whether the method is Scala 2 or not. Or assert it exists.

(We chose not to go this route for fear of false negatives.)

[smarter]

LGTM! (btw, once a PR is ready to review, you can assign a reviewer yourself so it doesn't get lost).