PR# 12878 Multi-constraint formal generics: calls on formal generic in watch expressions

Watch expressions that involve calls to features on entities of multi-constraint formal generic types result in unknown identifier (VEEN) or internal error.

1. Create and compile a system with the following classes:

	class A [G -> {
		B
			rename
				f as b_f,
				g as h,
				out as b_out
			end,
		C
			rename
				default_create as c_default_create,
				f as c_f,
				h as g,
				out as c_out
			end
		} create default_create end
	]
	feature
		f is
			do
				create item
			end
		item: G
	end

	class B
	inherit
		ANY
			redefine
				out
			end
	feature
		f, g, out: STRING is
			do
				Result := "B"
			end
	end

	class C
	inherit
		ANY
			redefine
				out
			end
	feature
		f, h, out: STRING is
			do
				Result := "C"
			end
	end

	class D
	inherit
		B
			rename
				f as b_f
			select
				out
			end
		C
			rename
				f as c_f,
				out as c_out
			end
	end

	class TEST
	create
		make
	feature
		make is
			local
				a: A [D]
			do
				create a
				a.f
			end
	end

where TEST is the root class of the system.

2. Run the system under debugger and stop in feature {A}.f after creation instruction.

3. Add the following "Watch" expressions:
	item.b_out
	item.c_out
	item.b_f
	item.c_f
	item.g
	item.h

4. Observe that the output does not match the expected result:
	B
	B
	B
	C
	C
	B

for now let's suspend this issue.
I'd like later to see if we can not do any thing more.

But I noticed the usual issue between runtime data and compiler data.
The runtime's data know the exacte generic derivation, the compiler knows only if this is a "reference".

`item' is of type D and not G (it is not formal).
That is why calls to `b_out' for example does NOT work as it does not occur in D.
`c_out' however actually occurs as a feature in D.
This might have caused some confusion. If you look at the flat view you will see that there is no more renaming visible.

While looking at it I found another bug and fixed it in rev#68897.

If you agree after double checking we might close this bug?