we need some test infrastructure that does exactly the same as
every single one of the test_pipe_caller.py unit tests, except
using ALUFunctionUnit, CRFunctionUnit etc. instead of
ALUBasePipe+ALUPipeSpec, CRBasePipe+CRPipeSpec etc.

the only pieces of code that need to change to achieve this is
how the inputs are set up, and the outputs obtained.

this basically means generalising each test_pipe_caller.py
TestRunner.run_all().process() function, so that instead
of explicitly targetting the alu instance:

                    vld = yield alu.n.valid_o
                    while not vld:
                        yield
                        vld = yield alu.n.valid_o
                    yield
                    alu_out = yield alu.n.data_o.o

instead it allows a parameterised selection of  the MultiCompUnit API /
data-synchronisation method (see experiment/compalu_multi.py tests):

    yield dut.issue_i.eq(1)
    yield
    yield dut.issue_i.eq(0)
    yield
    yield dut.rd.go.eq(0b11)
    while True:
        yield
        rd_rel_o = yield dut.rd.rel
        print ("rd_rel", rd_rel_o)
        if rd_rel_o:
            break

(actually, this ties in with Cesar's task, somewhat).