Use other tree for actual symbol of Assign

[som-snytt]

Fixes #22671

Find a name for the error message.

[som-snytt]

I started down the long tail of polishing words.

Let the symbol say what it is, and restrict setter explanation to class-owned symbols. (Need to check class params.)

[som-snytt]

Updated the branch from March, which had the spree contributors.

Improved to include the pathology from neg/assignments: x_= = 42. Some words for the user, and also don't try to format a message such as var x_=_= or similar, which throws in the formatter.

That test also suggested providing the args, which I took to mean show x_=(42).

I ran out of steam and did not add the scala 2 explanation, that x += 1 "does not convert to assignment" because the x is not mutable. The existing message that "it doesn't have a += member" is confusing, not only to beginners.

[odersky]

I think this PR would profit from a discussion on concrete test cases what the error message should be. Concentrate in common test cases not obscure ones. Obscure code does not warrant much work to make clear error messages.

So, can we break out the test files into a comment, show what the current message is, and what the suggested error message should be?

[odersky]

Sorry, I don't see how this is an improvement. Who in their right minds would write x_= =? That's just not what I see, and we should not special case it.

[som-snytt]

That was an existing test. And little cost to cover the case.

[odersky]

I find "Reassignment to val" much clearer than "Assignment to value".

[hamzaremmal]

I agree

[som-snytt]

Now the PR includes only the fix for the name issue.