in type inference at call sites, it is important to get the environment right

[dr2chase]

At least one sub-bug in getting generic functional methods right was the wrong choice of environment for inference of type parameters. This problem may exist for other cases of inference. We need to write tests to tickle this (something along the lines of A imports B, B imports C, and C defines things that get inferred at sites in A, with names chosen to overlap in A and C, to detect wrong choices.