Detect symbols inherited into package objects in ambiguity warning

[lrytz]

Symbols defined in package objects are not competing. The corresponding test failed to identify symbols inherited into package objects.

Fixes scala/bug#12816

Another follow-up for #10220

[som-snytt]

Very nice test!

I see in pos/t11921c, I added a test because of a Scaladoc test that exercises the code pattern where "nobody codes like that". But I suspect I did not think analytically about the case.

And I vaguely remember foundCompetingSymbol because of its unusual type (as a function), but not the details. I will never remember the isPackageXXX or hasPackageFlag API without an hour of research, like "cramming" for an exam and immediately forgetting whatever I may have learned.

I was about to say that I'm still not convinced that it's a false positive, while looking at your test, but from that previous pos test, I see that I must have decided the opposite back then. Who am I to argue with my past self?

My only annoying comment is that I prefer code comments to precede the block they document, so that it's easier to read the code as if the comment were folded (because usually I use vi). But it would be rude of me to prefer my quirk over your comfort level. I like the careting [sic], we should do that more often.

// look
if (b) case1
// look again
else case 2
//   ^^^^

[som-snytt]

You convinced me!

[som-snytt]

@lrytz thanks for updating the dotty PR. I see from the tests that that was where I changed my mind about this issue. Now, of course, I am agnostic, or possibly just apostate.